In [833]:
%matplotlib inline
import pandas as pd
import matplotlib.pyplot as plt
import datetime as dt  
import matplotlib
import numpy as np
import urllib
from scipy.stats import pearsonr

load seastar data


In [834]:
def loadalldatasetsmeasurements(list):
    measure = list[0]
    typeofmeasure = list[1]
    dataset =list[2]
    typelink= list[3]
    #change and remove the /test with real one
    if typeofmeasure=='b':
         result = pd.read_csv('http://141.26.208.201/datameasures/'+measure+typeofmeasure+'File_'+dataset+'_'+typelink+'.tsv',sep="\t",usecols=[0,1,2],dtype={'entity':str, 'typeentity':object, 'diffratio':float})
            #print('http://141.26.208.201/'+measure+typeofmeasure+'File_'+dataset+'_'+typelink+'.tsv')   
    elif typeofmeasure=='c':    
         result = pd.read_csv('http://141.26.208.201/datameasures/'+measure+typeofmeasure+'File_'+dataset+'_'+typelink+'.tsv',sep="\t",usecols=[0,1,2],dtype={'entity':str, 'typeentity':object, 'diffentropy':float})
            #print('http://141.26.208.201/'+measure+typeofmeasure+'File_'+dataset+'_'+typelink+'.tsv')
    result['dataset']=dataset
    return result

In [835]:
def loadalldatasets_ameasures(list):
    measure=list[0]
    dataset = list[1]
    typelink = list[2]
    if measure=='m11':
        result = pd.read_csv('http://141.26.208.201/datameasures/'+measure+'aFile_'+dataset+'_'+typelink+'.tsv',sep="\t",usecols=[0,1,2,3],dtype={'entity':str, 'typeentity':object, 'classifcardinality':float,'classifprimecardinality':float})
    elif measure=='m12':
        result = pd.read_csv('http://141.26.208.201/datameasures/'+measure+'aFile_'+dataset+'_'+typelink+'.tsv',sep="\t",usecols=[0,1,2,3],dtype={'entity':str, 'typeentity':object, 'desccardinality':float,'descprimecardinality':float})
    elif measure=='m13':
        result = pd.read_csv('http://141.26.208.201/datameasures/'+measure+'aFile_'+dataset+'_'+typelink+'.tsv',sep="\t",usecols=[0,1,2,3],dtype={'entity':str, 'typeentity':object, 'desccardinality':float,'descprimecardinality':float})
    elif measure=='m21':
        result = pd.read_csv('http://141.26.208.201/datameasures/'+measure+'aFile_'+dataset+'_'+typelink+'.tsv',sep="\t",usecols=[0,1,2,3],dtype={'entity':str, 'typeentity':object, 'econncardinality':float,'econnprimecardinality':float})
    elif measure=='m22':
        result = pd.read_csv('http://141.26.208.201/datameasures/'+measure+'aFile_'+dataset+'_'+typelink+'.tsv',sep="\t",usecols=[0,1,2,3],dtype={'entity':str, 'typeentity':object, 'datasetcardinality':float,'datasetprimecardinality':float})
    elif measure=='m31':
        result = pd.read_csv('http://141.26.208.201/datameasures/'+measure+'aFile_'+dataset+'_'+typelink+'.tsv',sep="\t",usecols=[0,1,2,3],dtype={'entity':str, 'typeentity':object, 'vocabdesccardinality':float,'vocabdescprimecardinality':float})
    result['dataset']=dataset
    return result

In [ ]:


In [ ]:


In [ ]:


In [836]:
datasets = ['l3s','sheffield','bne','bibbase','dwsmannheim','semanticweb','soton','dnb','bibsonomy','eurostat','openei','reegle','govwild','bfs','imf','fao','rodeionet','ndljp','icane','swcompany','korrekt','harth','revyu','eye48','simia','giftree','tomheath','twrpi','torre','linkedgeodata','aemet','geovocab','ontologi','ukpostcodes']

#datasets = ['l3s','sheffield','bne','bibbase','dwsmannheim','semanticweb','soton','dnb'] #tests


l2c_m11ai=[]
l2c_m12ai=[]
l2c_m13ai=[]
l2c_m21ai=[]
l2c_m22ai=[]
l2c_m31ai=[]

l2c_m11as=[]
l2c_m12as=[]
l2c_m13as=[]
l2c_m21as=[]
l2c_m22as=[]
l2c_m31as=[]

l2c_m11ar=[]
l2c_m12ar=[]
l2c_m13ar=[]
l2c_m21ar=[]
l2c_m22ar=[]
l2c_m31ar=[]

l2c_m11ao=[]
l2c_m12ao=[]
l2c_m13ao=[]
l2c_m21ao=[]
l2c_m22ao=[]
l2c_m31ao=[]

l2c_m11ac=[]
l2c_m12ac=[]
l2c_m13ac=[]
l2c_m21ac=[]
l2c_m22ac=[]
l2c_m31ac=[]

l2c_m11bi=[]
l2c_m12bi=[]
l2c_m13bi=[]
l2c_m21bi=[]
l2c_m22bi=[]
l2c_m31bi=[]

l2c_m11bs=[]
l2c_m12bs=[]
l2c_m13bs=[]
l2c_m21bs=[]
l2c_m22bs=[]
l2c_m31bs=[]

l2c_m11br=[]
l2c_m12br=[]
l2c_m13br=[]
l2c_m21br=[]
l2c_m22br=[]
l2c_m31br=[]

l2c_m11bo=[]
l2c_m12bo=[]
l2c_m13bo=[]
l2c_m21bo=[]
l2c_m22bo=[]
l2c_m31bo=[]

l2c_m11bc=[]
l2c_m12bc=[]
l2c_m13bc=[]
l2c_m21bc=[]
l2c_m22bc=[]
l2c_m31bc=[]

l2c_m11ci=[]
l2c_m12ci=[]
l2c_m13ci=[]
l2c_m21ci=[]
l2c_m22ci=[]
l2c_m31ci=[]

l2c_m11cs=[]
l2c_m12cs=[]
l2c_m13cs=[]
l2c_m21cs=[]
l2c_m22cs=[]
l2c_m31cs=[]

l2c_m11cr=[]
l2c_m12cr=[]
l2c_m13cr=[]
l2c_m21cr=[]
l2c_m22cr=[]
l2c_m31cr=[]

l2c_m11co=[]
l2c_m12co=[]
l2c_m13co=[]
l2c_m21co=[]
l2c_m22co=[]
l2c_m31co=[]

l2c_m11cc=[]
l2c_m12cc=[]
l2c_m13cc=[]
l2c_m21cc=[]
l2c_m22cc=[]
l2c_m31cc=[]


def getdata():
    for dataset in datasets:     
        ##### a 
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m11',dataset,'i'])
            l2c_m11ai.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m12',dataset,'i'])
            l2c_m12ai.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m13',dataset,'i'])
            l2c_m13ai.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m21',dataset,'i'])
            l2c_m12ai.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m22',dataset,'i'])
            l2c_m22ai.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m31',dataset,'i'])
            l2c_m31ai.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
       
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m11',dataset,'r'])
            l2c_m11ar.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m12',dataset,'r'])
            l2c_m12ar.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m13',dataset,'r'])
            l2c_m13ar.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m21',dataset,'r'])
            l2c_m12ar.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m22',dataset,'r'])
            l2c_m22ar.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m31',dataset,'r'])
            l2c_m31ar.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m11',dataset,'s'])
            l2c_m11as.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m12',dataset,'s'])
            l2c_m12as.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m13',dataset,'s'])
            l2c_m13as.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m21',dataset,'s'])
            l2c_m12as.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m22',dataset,'s'])
            l2c_m22as.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m31',dataset,'s'])
            l2c_m31as.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m11',dataset,'o'])
            l2c_m11ao.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m12',dataset,'o'])
            l2c_m12ao.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m13',dataset,'o'])
            l2c_m13ao.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m21',dataset,'o'])
            l2c_m12ao.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m22',dataset,'o'])
            l2c_m22ao.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m31',dataset,'o'])
            l2c_m31ao.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
       
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m11',dataset,'c'])
            l2c_m11ac.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m12',dataset,'c'])
            l2c_m12ac.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m13',dataset,'c'])
            l2c_m13ac.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m21',dataset,'c'])
            l2c_m12ac.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m22',dataset,'c'])
            l2c_m22ac.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:
            #all measurements for links type='i'
            measurement=loadalldatasets_ameasures(['m31',dataset,'c'])
            l2c_m31ac.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
       
        ##### b and c
        try:
            #all measurements for links type='i'
            measurement=loadalldatasetsmeasurements(['m11','b',dataset,'i'])
            #print(dataset)
            #print(measurement)
            l2c_m11bi.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:    
            measurement=loadalldatasetsmeasurements(['m12','b',dataset,'i'])
            #print(dataset)
            #print(measurement)
            l2c_m12bi.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:    
            measurement=loadalldatasetsmeasurements(['m13','b',dataset,'i'])
            l2c_m13bi.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:    
            measurement=loadalldatasetsmeasurements(['m21','b',dataset,'i'])
            l2c_m21bi.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:    
            measurement=loadalldatasetsmeasurements(['m22','b',dataset,'i'])
            l2c_m22bi.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)
        try:        
            measurement=loadalldatasetsmeasurements(['m31','b',dataset,'i'])
            l2c_m31bi.append(measurement) 
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:   
            #all measurements for links type='s'
            measurement=loadalldatasetsmeasurements(['m11','b',dataset,'s'])
            l2c_m11bs.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:    
            measurement=loadalldatasetsmeasurements(['m12','b',dataset,'s'])
            l2c_m12bs.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:    
            measurement=loadalldatasetsmeasurements(['m13','b',dataset,'s'])
            l2c_m13bs.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:    
            measurement=loadalldatasetsmeasurements(['m21','b',dataset,'s'])
            l2c_m21bs.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:    
            measurement=loadalldatasetsmeasurements(['m22','b',dataset,'s'])
            l2c_m22bs.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:    
            measurement=loadalldatasetsmeasurements(['m31','b',dataset,'s'])
            l2c_m31bs.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
            #all measurements for links type='r'
            
        try:    
            measurement=loadalldatasetsmeasurements(['m11','b',dataset,'r'])
            l2c_m11br.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:    
            measurement=loadalldatasetsmeasurements(['m12','b',dataset,'r'])
            l2c_m12br.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)
        try:       
            measurement=loadalldatasetsmeasurements(['m13','b',dataset,'r'])
            l2c_m13br.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try: 
            measurement=loadalldatasetsmeasurements(['m21','b',dataset,'r'])
            l2c_m21br.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:    
            measurement=loadalldatasetsmeasurements(['m22','b',dataset,'r'])
            l2c_m22br.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:   
            measurement=loadalldatasetsmeasurements(['m31','b',dataset,'r'])
            l2c_m31br.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)
        try:        
            #all measurements for links type='o'
            measurement=loadalldatasetsmeasurements(['m11','b',dataset,'o'])
            l2c_m11bo.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)   
        try:   
            measurement=loadalldatasetsmeasurements(['m12','b',dataset,'o'])
            l2c_m12bo.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:    
            measurement=loadalldatasetsmeasurements(['m13','b',dataset,'o'])
            l2c_m13bo.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:    
            measurement=loadalldatasetsmeasurements(['m21','b',dataset,'o'])
            l2c_m21bo.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)       
        try:    
            measurement=loadalldatasetsmeasurements(['m22','b',dataset,'o'])
            l2c_m22bo.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:    
            measurement=loadalldatasetsmeasurements(['m31','b',dataset,'o'])
            l2c_m31bo.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
            #all measurements for links type='c'
        try:    
            measurement=loadalldatasetsmeasurements(['m11','b',dataset,'c'])
            l2c_m11bc.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:    
            measurement=loadalldatasetsmeasurements(['m12','b',dataset,'c'])
            l2c_m12bc.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:   
            measurement=loadalldatasetsmeasurements(['m13','b',dataset,'c'])
            l2c_m13bc.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:    
            measurement=loadalldatasetsmeasurements(['m21','b',dataset,'c'])
            l2c_m21bc.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:    
            measurement=loadalldatasetsmeasurements(['m22','b',dataset,'c'])
            l2c_m22bc.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:    
            measurement=loadalldatasetsmeasurements(['m31','b',dataset,'c'])
            l2c_m31bc.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
            #listtoconcatenate l2c -------------------------
        try:   
            measurement=loadalldatasetsmeasurements(['m11','c',dataset,'i'])
            l2c_m11ci.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:    
            measurement=loadalldatasetsmeasurements(['m12','c',dataset,'i'])
            l2c_m12ci.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:    
            measurement=loadalldatasetsmeasurements(['m13','c',dataset,'i'])
            l2c_m13ci.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:    
            measurement=loadalldatasetsmeasurements(['m21','c',dataset,'i'])
            l2c_m21ci.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:        
            measurement=loadalldatasetsmeasurements(['m22','c',dataset,'i'])
            l2c_m22ci.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)
        try:        
            measurement=loadalldatasetsmeasurements(['m31','c',dataset,'i'])
            l2c_m31ci.append(measurement) 
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)
        try:        
            #all measurements for links type='s'
            measurement=loadalldatasetsmeasurements(['m11','c',dataset,'s'])
            l2c_m11cs.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)
        try:        
            measurement=loadalldatasetsmeasurements(['m12','c',dataset,'s'])
            l2c_m12cs.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:    
            measurement=loadalldatasetsmeasurements(['m13','c',dataset,'s'])
            l2c_m13cs.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)
        try:        
            measurement=loadalldatasetsmeasurements(['m21','c',dataset,'s'])
            l2c_m21cs.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:    
            measurement=loadalldatasetsmeasurements(['m22','c',dataset,'s'])
            l2c_m22cs.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)
        try:        
            measurement=loadalldatasetsmeasurements(['m31','c',dataset,'s'])
            l2c_m31cs.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
            #all measurements for links type='r'
        try:    
            measurement=loadalldatasetsmeasurements(['m11','c',dataset,'r'])
            l2c_m11cr.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)
        try:        
            measurement=loadalldatasetsmeasurements(['m12','c',dataset,'r'])
            l2c_m12cr.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:    
            measurement=loadalldatasetsmeasurements(['m13','c',dataset,'r'])
            l2c_m13cr.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)
        try:        
            measurement=loadalldatasetsmeasurements(['m21','c',dataset,'r'])
            l2c_m21cr.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)
        try:        
            measurement=loadalldatasetsmeasurements(['m22','c',dataset,'r'])
            l2c_m22cr.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)
        try:        
            measurement=loadalldatasetsmeasurements(['m31','c',dataset,'r'])
            l2c_m31cr.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)
        try:        
            #all measurements for links type='o'
            measurement=loadalldatasetsmeasurements(['m11','c',dataset,'o'])
            l2c_m11co.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)
        try:        
            measurement=loadalldatasetsmeasurements(['m12','c',dataset,'o'])
            l2c_m12co.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)
        try:        
            measurement=loadalldatasetsmeasurements(['m13','c',dataset,'o'])
            l2c_m13co.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)    
        try:    
            measurement=loadalldatasetsmeasurements(['m21','c',dataset,'o'])
            l2c_m21co.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)
        try:        
            measurement=loadalldatasetsmeasurements(['m22','c',dataset,'o'])
            l2c_m22co.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)
        try:        
            measurement=loadalldatasetsmeasurements(['m31','c',dataset,'o'])
            l2c_m31co.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)
        try:        
            #all measurements for links type='c'
            measurement=loadalldatasetsmeasurements(['m11','c',dataset,'c'])
            l2c_m11cc.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)
        try:        
            measurement=loadalldatasetsmeasurements(['m12','c',dataset,'c'])
            l2c_m12cc.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)
        try:        
            measurement=loadalldatasetsmeasurements(['m13','c',dataset,'c'])
            l2c_m13cc.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)  
        try:        
            measurement=loadalldatasetsmeasurements(['m21','c',dataset,'c'])
            l2c_m21cc.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)
        try:        
            measurement=loadalldatasetsmeasurements(['m22','c',dataset,'c'])
            l2c_m22cc.append(measurement)
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)
        try:        
            measurement=loadalldatasetsmeasurements(['m31','c',dataset,'c'])
            l2c_m31cc.append(measurement)
            #listtoconcatenate l2c
        except urllib.error.HTTPError as err:
            if err.code == 404:
                print('HTTP error when loading all measurements for a measure for data set: '+dataset)
                          
getdata()


HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: dwsmannheim
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: bibsonomy
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: openei
HTTP error when loading all measurements for a measure for data set: reegle
HTTP error when loading all measurements for a measure for data set: reegle
HTTP error when loading all measurements for a measure for data set: reegle
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: govwild
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: ndljp
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: swcompany
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: giftree
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre
HTTP error when loading all measurements for a measure for data set: torre

In [837]:
print(str(len(l2c_m22br))) # still accesible after modified by the function (global var)


26

In [838]:
def ratioandentropyinonetable(list):
    concat = pd.concat(list, axis=1)
    return concat

In [ ]:


In [839]:
def inequalities(b,c):
    ratioandentropy=[]
    ratioandentropy.append(b)
    ratioandentropy.append(c['diffentropy']) # the columns entity, typeentity,dataset are the same, just add the column on diffentropy
    bc = ratioandentropyinonetable(ratioandentropy)
    bc.head(3)
    #bc2 = bc.drop(bc.columns[[3, 5]], axis=1)
    
    #print(bc.head(5))
    
    #groupedbc = bc.groupby('dataset')
    #print(groupedbc['diffentropy'].describe())
    
    #print(groupedbc['diffratio'].describe())
    bc_grouped =bc.groupby('dataset')
    bc_grouped.boxplot(column='diffentropy',figsize=(20,20))
    #bc_grouped.boxplot(column=['diffratio','diffentropy'],figsize=(20,20)) #GOOD for showing both ratio and entropy
    #bc.boxplot(column='diffratio',by='dataset')
    #bc.boxplot(column='diffentropy',by='dataset')

In [840]:
concat_m11ai=pd.DataFrame()
concat_m12ai=pd.DataFrame()
concat_m13ai=pd.DataFrame()
concat_m21ai=pd.DataFrame()
concat_m22ai=pd.DataFrame()
concat_m31ai=pd.DataFrame()

concat_m11as=pd.DataFrame()
concat_m12as=pd.DataFrame()
concat_m13as=pd.DataFrame()
concat_m21as=pd.DataFrame()
concat_m22as=pd.DataFrame()
concat_m31as=pd.DataFrame()

concat_m11ar=pd.DataFrame()
concat_m12ar=pd.DataFrame()
concat_m13ar=pd.DataFrame()
concat_m21ar=pd.DataFrame()
concat_m22ar=pd.DataFrame()
concat_m31ar=pd.DataFrame()

concat_m11ao=pd.DataFrame()
concat_m12ao=pd.DataFrame()
concat_m13ao=pd.DataFrame()
concat_m21ao=pd.DataFrame()
concat_m22ao=pd.DataFrame()
concat_m31ao=pd.DataFrame()

concat_m11ac=pd.DataFrame()
concat_m12ac=pd.DataFrame()
concat_m13ac=pd.DataFrame()
concat_m21ac=pd.DataFrame()
concat_m22ac=pd.DataFrame()
concat_m31ac=pd.DataFrame()

In [841]:
if len(l2c_m11ai) != 0:
    concat_m11ai =pd.concat(l2c_m11ai)
if len(l2c_m12ai) != 0:
    concat_m12ai =pd.concat(l2c_m12ai)
if len(l2c_m13ai) != 0:
    concat_m13ai =pd.concat(l2c_m13ai)
if len(l2c_m21ai) != 0:
    concat_m21ai =pd.concat(l2c_m21ai)
if len(l2c_m22ai) != 0:
    concat_m22ai =pd.concat(l2c_m22ai)
if len(l2c_m31ai) != 0:
    concat_m31ai =pd.concat(l2c_m31ai)
    
if len(l2c_m11as) != 0:
    concat_m11as =pd.concat(l2c_m11as)
if len(l2c_m12as) != 0:
    concat_m12as =pd.concat(l2c_m12as)
if len(l2c_m13as) != 0:
    concat_m13as =pd.concat(l2c_m13as)
if len(l2c_m21as) != 0:
    concat_m21as =pd.concat(l2c_m21as)
if len(l2c_m22as) != 0:
    concat_m22as =pd.concat(l2c_m22as)
if len(l2c_m31as) != 0:
    concat_m31as =pd.concat(l2c_m31as)
    
if len(l2c_m11ar) != 0:
    concat_m11ar =pd.concat(l2c_m11ar)
if len(l2c_m12ar) != 0:
    concat_m12ar =pd.concat(l2c_m12ar)
if len(l2c_m13ar) != 0:
    concat_m13ar =pd.concat(l2c_m13ar)
if len(l2c_m21ar) != 0:
    concat_m21ar =pd.concat(l2c_m21ar)
if len(l2c_m22ar) != 0:
    concat_m22ar =pd.concat(l2c_m22ar)
if len(l2c_m31ar) != 0:
    concat_m31ar =pd.concat(l2c_m31ar)

if len(l2c_m11ao) != 0:
    concat_m11ao =pd.concat(l2c_m11ao)
if len(l2c_m12ao) != 0:
    concat_m12ao =pd.concat(l2c_m12ao)
if len(l2c_m13ao) != 0:
    concat_m13ao =pd.concat(l2c_m13ao)
if len(l2c_m21ao) != 0:
    concat_m21ao =pd.concat(l2c_m21ao)
if len(l2c_m22ao) != 0:
    concat_m22ao =pd.concat(l2c_m22ao)
if len(l2c_m31ao) != 0:
    concat_m31ao =pd.concat(l2c_m31ao)

if len(l2c_m11ac) != 0:
    concat_m11ac =pd.concat(l2c_m11ac)
if len(l2c_m12ac) != 0:
    concat_m12ac =pd.concat(l2c_m12ac)
if len(l2c_m13ac) != 0:
    concat_m13ac =pd.concat(l2c_m13ac)
if len(l2c_m21ac) != 0:
    concat_m21ac =pd.concat(l2c_m21ac)
if len(l2c_m22ac) != 0:
    concat_m22ac =pd.concat(l2c_m22ac)
if len(l2c_m31ac) != 0:
    concat_m31ac =pd.concat(l2c_m31ac)

    
print(concat_m11ai.head(2))


                                              entity typeentity  \
0  http://dblp.l3s.de/d2r/resource/publications/c...       null   
1  http://dblp.l3s.de/d2r/resource/publications/c...       null   

   classifcardinality  classifprimecardinality dataset  
0                 0.0                      0.0     l3s  
1                 0.0                      0.0     l3s  

In [842]:
#####

concat_m11bi=pd.DataFrame()
concat_m12bi=pd.DataFrame()
concat_m13bi=pd.DataFrame()
concat_m21bi=pd.DataFrame()
concat_m22bi=pd.DataFrame()
concat_m31bi=pd.DataFrame()

concat_m11ci=pd.DataFrame()
concat_m12ci=pd.DataFrame()
concat_m13ci=pd.DataFrame()
concat_m21ci=pd.DataFrame()
concat_m22ci=pd.DataFrame()
concat_m31ci=pd.DataFrame()

concat_m11bs=pd.DataFrame()
concat_m12bs=pd.DataFrame()
concat_m13bs=pd.DataFrame()
concat_m21bs=pd.DataFrame()
concat_m22bs=pd.DataFrame()
concat_m31bs=pd.DataFrame()

concat_m11cs=pd.DataFrame()
concat_m12cs=pd.DataFrame()
concat_m13cs=pd.DataFrame()
concat_m21cs=pd.DataFrame()
concat_m22cs=pd.DataFrame()
concat_m31cs=pd.DataFrame()

concat_m11br=pd.DataFrame()
concat_m12br=pd.DataFrame()
concat_m13br=pd.DataFrame()
concat_m21br=pd.DataFrame()
concat_m22br=pd.DataFrame()
concat_m31br=pd.DataFrame()

concat_m11cr=pd.DataFrame()
concat_m12cr=pd.DataFrame()
concat_m13cr=pd.DataFrame()
concat_m21cr=pd.DataFrame()
concat_m22cr=pd.DataFrame()
concat_m31cr=pd.DataFrame()

concat_m11bo=pd.DataFrame()
concat_m12bo=pd.DataFrame()
concat_m13bo=pd.DataFrame()
concat_m21bo=pd.DataFrame()
concat_m22bo=pd.DataFrame()
concat_m31bo=pd.DataFrame()

concat_m11co=pd.DataFrame()
concat_m12co=pd.DataFrame()
concat_m13co=pd.DataFrame()
concat_m21co=pd.DataFrame()
concat_m22co=pd.DataFrame()
concat_m31co=pd.DataFrame()

concat_m11bc=pd.DataFrame()
concat_m12bc=pd.DataFrame()
concat_m13bc=pd.DataFrame()
concat_m21bc=pd.DataFrame()
concat_m22bc=pd.DataFrame()
concat_m31bc=pd.DataFrame()

concat_m11cc=pd.DataFrame()
concat_m12cc=pd.DataFrame()
concat_m13cc=pd.DataFrame()
concat_m21cc=pd.DataFrame()
concat_m22cc=pd.DataFrame()
concat_m31cc=pd.DataFrame()

In [843]:
if len(l2c_m11bi) != 0 and len(l2c_m11ci)!= 0:
    print('m11 i check')
    concat_m11bi =pd.concat(l2c_m11bi)
    concat_m11ci = pd.concat(l2c_m11ci)
    inequalities(concat_m11bi, concat_m11ci)

if len(l2c_m12bi) != 0 and len(l2c_m12ci)!= 0:
    print('m12 i check')
    concat_m12bi =pd.concat(l2c_m12bi)
    concat_m12ci = pd.concat(l2c_m12ci)
    inequalities(concat_m12bi, concat_m12ci)

if len(l2c_m13bi) != 0 and len(l2c_m13ci)!= 0:
    print('m13 i check')
    concat_m13bi =pd.concat(l2c_m13bi)
    concat_m13ci = pd.concat(l2c_m13ci)
    inequalities(concat_m13bi, concat_m13ci)

if len(l2c_m21bi) != 0 and len(l2c_m21ci)!= 0:
    print('m21 i check')
    concat_m21bi =pd.concat(l2c_m21bi)
    concat_m21ci = pd.concat(l2c_m21ci)
    inequalities(concat_m21bi, concat_m21ci)

if len(l2c_m22bi) != 0 and len(l2c_m22ci)!= 0:
    print('m22 i check')
    concat_m22bi =pd.concat(l2c_m22bi)
    concat_m22ci = pd.concat(l2c_m22ci)
    inequalities(concat_m22bi, concat_m22ci)

if len(l2c_m31bi) != 0 and len(l2c_m31ci)!= 0:
    print('m31 i check')
    concat_m31bi =pd.concat(l2c_m31bi)
    concat_m31ci = pd.concat(l2c_m31ci)
    inequalities(concat_m31bi, concat_m31ci)


m11 i check
m12 i check
m13 i check
m21 i check
m22 i check
m31 i check

In [857]:
if len(l2c_m11bs) != 0 and len(l2c_m11cs)!= 0:
    concat_m11bs =pd.concat(l2c_m11bs)
    concat_m11cs = pd.concat(l2c_m11cs)
    inequalities(concat_m11bs, concat_m11cs)

if len(l2c_m12bs) != 0 and len(l2c_m12cs)!= 0:
    concat_m12bs =pd.concat(l2c_m12bs)
    concat_m12cs = pd.concat(l2c_m12cs)
    inequalities(concat_m12bs, concat_m12cs)

if len(l2c_m13bs) != 0 and len(l2c_m13cs)!= 0:
    concat_m13bs =pd.concat(l2c_m13bs)
    concat_m13cs = pd.concat(l2c_m13cs)
    inequalities(concat_m13bs, concat_m13cs)

if len(l2c_m21bs) != 0 and len(l2c_m21cs)!= 0:
    concat_m21bs =pd.concat(l2c_m21bs)
    concat_m21cs = pd.concat(l2c_m21cs)
    inequalities(concat_m21bs, concat_m21cs)

if len(l2c_m22bs) != 0 and len(l2c_m22cs)!= 0:
    concat_m22bs =pd.concat(l2c_m22bs)
    concat_m22cs = pd.concat(l2c_m22cs)
    inequalities(concat_m22bs, concat_m22cs)

if len(l2c_m31bs) != 0 and len(l2c_m31cs)!= 0:
    concat_m31bs =pd.concat(l2c_m31bs)
    concat_m31cs = pd.concat(l2c_m31cs)
    inequalities(concat_m31bs, concat_m31cs)



In [859]:
if len(l2c_m11br) != 0 and len(l2c_m11cr)!= 0:
    concat_m11br =pd.concat(l2c_m11br)
    concat_m11cr = pd.concat(l2c_m11cr)
    inequalities(concat_m11br, concat_m11cr)
    
if len(l2c_m12br) != 0 and len(l2c_m12cr)!= 0:
    concat_m12br =pd.concat(l2c_m12br)
    concat_m12cr = pd.concat(l2c_m12cr)
    inequalities(concat_m12br, concat_m12cr)

if len(l2c_m13br) != 0 and len(l2c_m13cr)!= 0:
    concat_m13br =pd.concat(l2c_m13br)
    concat_m13cr = pd.concat(l2c_m13cr)
    inequalities(concat_m13br, concat_m13cr)

if len(l2c_m21br) != 0 and len(l2c_m21cr)!= 0:
    concat_m21br =pd.concat(l2c_m21br)
    concat_m21cr = pd.concat(l2c_m21cr)
    inequalities(concat_m21br, concat_m21cr)

if len(l2c_m22br) != 0 and len(l2c_m22cr)!= 0:
    concat_m22br =pd.concat(l2c_m22br)
    concat_m22cr = pd.concat(l2c_m22cr)
    inequalities(concat_m22br, concat_m22cr)

if len(l2c_m31br) != 0 and len(l2c_m31cr)!= 0:
    concat_m31br =pd.concat(l2c_m31br)
    concat_m31cr = pd.concat(l2c_m31cr)
    inequalities(concat_m31br, concat_m31cr)



In [860]:
if len(l2c_m11bo) != 0 and len(l2c_m11co)!= 0:
    concat_m11bo =pd.concat(l2c_m11bo)
    concat_m11co = pd.concat(l2c_m11co)
    inequalities(concat_m11bo, concat_m11co)

if len(l2c_m12bo) != 0 and len(l2c_m12co)!= 0:
    concat_m12bo =pd.concat(l2c_m12bo)
    concat_m12co = pd.concat(l2c_m12co)
    inequalities(concat_m12bo, concat_m12co)

if len(l2c_m13bo) != 0 and len(l2c_m13co)!= 0:
    concat_m13bo =pd.concat(l2c_m13bo)
    concat_m13co = pd.concat(l2c_m13co)
    inequalities(concat_m13bo, concat_m13co)

if len(l2c_m21bo) != 0 and len(l2c_m21co)!= 0:
    concat_m21bo =pd.concat(l2c_m21bo)
    concat_m21co = pd.concat(l2c_m21co)
    inequalities(concat_m21bo, concat_m21co)

if len(l2c_m22bo) != 0 and len(l2c_m22co)!= 0: 
    concat_m22bo =pd.concat(l2c_m22bo)
    concat_m22co = pd.concat(l2c_m22co)
    inequalities(concat_m22bo, concat_m22co)

if len(l2c_m31bo) != 0 and len(l2c_m31co)!= 0:
    concat_m31bo =pd.concat(l2c_m31bo)
    concat_m31co = pd.concat(l2c_m31co)
    inequalities(concat_m31bo, concat_m31co)



In [861]:
if len(l2c_m11bc) != 0 and len(l2c_m11cc)!= 0:
    concat_m11bc =pd.concat(l2c_m11bc)
    concat_m11cc = pd.concat(l2c_m11cc)
    inequalities(concat_m11bc, concat_m11cc)

if len(l2c_m12bc) != 0 and len(l2c_m12cc)!= 0:
    concat_m12bc =pd.concat(l2c_m12bc)
    concat_m12cc = pd.concat(l2c_m12cc)
    inequalities(concat_m12bc, concat_m12cc)

    
if len(l2c_m13bc) != 0 and len(l2c_m13cc)!= 0:
    concat_m13bc =pd.concat(l2c_m13bc)
    concat_m13cc = pd.concat(l2c_m13cc)
    inequalities(concat_m13bc, concat_m13cc)


if len(l2c_m21bc) != 0 and len(l2c_m21cc)!= 0:
    concat_m21bc =pd.concat(l2c_m21bc)
    concat_m21cc = pd.concat(l2c_m21cc)
    inequalities(concat_m21bc, concat_m21cc)


if len(l2c_m22bc) != 0 and len(l2c_m22cc)!= 0:
    concat_m22bc =pd.concat(l2c_m22bc)
    concat_m22cc = pd.concat(l2c_m22cc)
    inequalities(concat_m22bc, concat_m22cc)


if len(l2c_m31bc) != 0 and len(l2c_m31cc)!= 0:
    concat_m31bc =pd.concat(l2c_m31bc)
    concat_m31cc = pd.concat(l2c_m31cc)
    inequalities(concat_m31bc, concat_m31cc)


discriminative


In [862]:
def describe_discriminative(list):
    df= list[0]
    mi=list[1]
    if len(df)!=0:
        if 'diffentropy' in df:
            df.groupby('dataset').diffentropy.describe().to_csv(mi, header=True) 
            print(df.groupby('dataset').diffentropy.mean())
        elif 'diffratio' in df:
            df.groupby('dataset').diffratio.describe().to_csv(mi, header=True)
            print(df.groupby('dataset').diffratio.mean())
        else: 
            df.groupby('dataset').describe().to_csv(mi, header=True)
            print(df.groupby('dataset').mean())

In [863]:
with open('m11a_alltypesoflinks_describe.csv', 'a') as mi:
    
    concat_m11a = pd.concat([concat_m11ai,concat_m11as,concat_m11ar,concat_m11ao,concat_m11ac])
    describe_discriminative([concat_m11a,mi])
    print('m11a')
    
with open('m11b_alltypesoflinks_describe.csv', 'a') as mi:
    concat_m11b = pd.concat([concat_m11bi,concat_m11bs,concat_m11br,concat_m11bo,concat_m11bc])
    describe_discriminative([concat_m11b,mi])
    print('m11b')
    
with open('m11c_alltypesoflinks_describe.csv', 'a') as mi:
    concat_m11c = pd.concat([concat_m11ci,concat_m11cs,concat_m11cr,concat_m11co,concat_m11cc])
    describe_discriminative([concat_m11a,mi])
    print('m11c')

with open('m12a_alltypesoflinks_describe.csv', 'a') as mi:
    concat_m12a = pd.concat([concat_m12ai,concat_m12as,concat_m12ar,concat_m12ao,concat_m12ac])
    describe_discriminative([concat_m12a,mi])
    print('m12a')
with open('m12b_alltypesoflinks_describe.csv', 'a') as mi:
    concat_m12b = pd.concat([concat_m12bi,concat_m12bs,concat_m12br,concat_m12bo,concat_m12bc])
    describe_discriminative([concat_m12b,mi])
    print('m12b')
with open('m12c_alltypesoflinks_describe.csv', 'a') as mi:
    concat_m12c = pd.concat([concat_m12ci,concat_m12cs,concat_m12cr,concat_m12co,concat_m12cc])
    describe_discriminative([concat_m12c,mi]) 
    print('m12c')
with open('m13a_alltypesoflinks_describe.csv', 'a') as mi:
    concat_m13a = pd.concat([concat_m13ai,concat_m13as,concat_m13ar,concat_m13ao,concat_m13ac])
    describe_discriminative([concat_m13a,mi])
    print('m13a')
with open('m13b_alltypesoflinks_describe.csv', 'a') as mi:
    concat_m13b = pd.concat([concat_m13bi,concat_m13bs,concat_m13br,concat_m13bo,concat_m13bc])
    describe_discriminative([concat_m13b,mi])
    print('m13b')
with open('m13c_alltypesoflinks_describe.csv', 'a') as mi:
    concat_m13c = pd.concat([concat_m13ci,concat_m13cs,concat_m13cr,concat_m13co,concat_m13cc])
    describe_discriminative([concat_m13c,mi]) 
    print('m13c')
    
with open('m21a_alltypesoflinks_describe.csv', 'a') as mi:
    concat_m21a = pd.concat([concat_m21ai,concat_m21as,concat_m21ar,concat_m21ao,concat_m21ac])
    describe_discriminative([concat_m21a,mi])
    print('m21a')
with open('m21b_alltypesoflinks_describe.csv', 'a') as mi:
    concat_m21b = pd.concat([concat_m21bi,concat_m21bs,concat_m21br,concat_m21bo,concat_m21bc])
    describe_discriminative([concat_m21b,mi])
    print('m21b')
with open('m21c_alltypesoflinks_describe.csv', 'a') as mi:
    concat_m21c = pd.concat([concat_m21ci,concat_m21cs,concat_m21cr,concat_m21co,concat_m21cc])
    describe_discriminative([concat_m21c,mi])    
    print('m21c')
with open('m22a_alltypesoflinks_describe.csv', 'a') as mi:
    concat_m22a = pd.concat([concat_m22ai,concat_m22as,concat_m22ar,concat_m22ao,concat_m22ac])
    describe_discriminative([concat_m22a,mi])
    print('m22a')
with open('m22b_alltypesoflinks_describe.csv', 'a') as mi:
    concat_m22b = pd.concat([concat_m22bi,concat_m22bs,concat_m22br,concat_m22bo,concat_m22bc])
    describe_discriminative([concat_m22b,mi])
    print('m22b')
with open('m22c_alltypesoflinks_describe.csv', 'a') as mi:
    concat_m22c = pd.concat([concat_m22ci,concat_m22cs,concat_m22cr,concat_m22co,concat_m22cc])
    describe_discriminative([concat_m22c,mi])  
    print('m22c')

with open('m31a_alltypesoflinks_describe.csv', 'a') as mi:
    concat_m31a = pd.concat([concat_m31ai,concat_m31as,concat_m31ar,concat_m31ao,concat_m31ac])
    describe_discriminative([concat_m31a,mi])   
    print('m31a')
with open('m31b_alltypesoflinks_describe.csv', 'a') as mi:
    concat_m31b = pd.concat([concat_m31bi,concat_m31bs,concat_m31br,concat_m31bo,concat_m31bc])
    describe_discriminative([concat_m31b,mi])   
    print('m31b')
with open('m31c_alltypesoflinks_describe.csv', 'a') as mi:
    concat_m31c = pd.concat([concat_m31ci,concat_m31cs,concat_m31cr,concat_m31co,concat_m31cc])
    describe_discriminative([concat_m31c,mi])  
    print('m31c')


               classifcardinality  classifprimecardinality
dataset                                                   
aemet                    0.022556                 0.022556
bfs                      0.312198                 0.312198
bibbase                  0.995482                 0.995482
bibsonomy                0.000000                 0.000000
bne                      0.000000                 0.000000
dnb                      0.611729                 0.611729
dwsmannheim              0.000000                 0.000000
eurostat                 0.479550                 0.479550
eye48                    0.000000                 0.000000
fao                      0.000000                 0.000000
geovocab                 0.000000                 0.000000
govwild                  0.000000                 0.000000
harth                    0.000000                 0.000000
icane                    0.545455                 0.545455
imf                      0.306452                 0.306452
korrekt                  0.566806                 0.566806
l3s                      0.000000                 0.000000
linkedgeodata            2.804388                 2.804388
ndljp                    0.000000                 0.000000
ontologi                 0.223301                 0.223301
openei                   3.000000                 3.000000
reegle                   0.000000                 0.000000
revyu                    0.000000                 0.000000
rodeionet                0.995208                 0.995208
semanticweb              1.054148                 1.054148
sheffield                0.000000                 0.000000
simia                    0.896885                 0.896885
soton                    0.000000                 0.000000
swcompany                0.000000                 0.000000
tomheath                 0.000000                 0.000000
twrpi                    0.000000                 0.000000
ukpostcodes              0.000000                 0.000000
m11a
dataset
aemet            0.0
bfs              0.0
bibbase          0.0
bibsonomy        0.0
bne              0.0
dnb              0.0
dwsmannheim      0.0
eurostat         0.0
eye48            0.0
fao              0.0
geovocab         0.0
govwild          0.0
harth            0.0
icane            0.0
imf              0.0
korrekt          0.0
l3s              0.0
linkedgeodata    0.0
ndljp            0.0
ontologi         0.0
openei           0.0
reegle           0.0
revyu            0.0
rodeionet        0.0
semanticweb      0.0
sheffield        0.0
simia            0.0
soton            0.0
swcompany        0.0
tomheath         0.0
twrpi            0.0
ukpostcodes      0.0
Name: diffratio, dtype: float64
m11b
               classifcardinality  classifprimecardinality
dataset                                                   
aemet                    0.022556                 0.022556
bfs                      0.312198                 0.312198
bibbase                  0.995482                 0.995482
bibsonomy                0.000000                 0.000000
bne                      0.000000                 0.000000
dnb                      0.611729                 0.611729
dwsmannheim              0.000000                 0.000000
eurostat                 0.479550                 0.479550
eye48                    0.000000                 0.000000
fao                      0.000000                 0.000000
geovocab                 0.000000                 0.000000
govwild                  0.000000                 0.000000
harth                    0.000000                 0.000000
icane                    0.545455                 0.545455
imf                      0.306452                 0.306452
korrekt                  0.566806                 0.566806
l3s                      0.000000                 0.000000
linkedgeodata            2.804388                 2.804388
ndljp                    0.000000                 0.000000
ontologi                 0.223301                 0.223301
openei                   3.000000                 3.000000
reegle                   0.000000                 0.000000
revyu                    0.000000                 0.000000
rodeionet                0.995208                 0.995208
semanticweb              1.054148                 1.054148
sheffield                0.000000                 0.000000
simia                    0.896885                 0.896885
soton                    0.000000                 0.000000
swcompany                0.000000                 0.000000
tomheath                 0.000000                 0.000000
twrpi                    0.000000                 0.000000
ukpostcodes              0.000000                 0.000000
m11c
               desccardinality  descprimecardinality  econncardinality  \
dataset                                                                  
aemet                 3.819549              4.969925          0.714286   
bfs                   5.909726             41.919936          2.930145   
bibbase               8.774096             11.143072          4.579819   
bibsonomy            16.666554             16.723321          1.991834   
bne                   9.840741             11.170370          7.274074   
dnb                   8.581171             18.688911          4.118414   
dwsmannheim           4.732394             10.042254          3.323944   
eurostat             12.873211             36.865031          2.422290   
eye48                 0.000000              6.000000          0.000000   
fao                  10.400000             12.333333          4.533333   
geovocab              0.000000              7.445566          0.000000   
govwild               0.000000              0.000000          0.000000   
harth                 3.482759             18.931034          1.924138   
icane                 2.272727             32.454545          0.454545   
imf                   4.072581             41.254032          1.814516   
korrekt              13.592036             14.591707          8.374177   
l3s                  11.458394             13.119392          4.153401   
linkedgeodata        68.531993            184.685558          1.882998   
ndljp                 7.000000              7.000000          2.000000   
ontologi            193.788648            198.513816        185.489171   
openei               44.250000             44.250000         25.000000   
reegle               10.855967            137.427984          4.888889   
revyu                 4.154209              5.468254          2.976976   
rodeionet            48.982428             50.800319         32.864217   
semanticweb          19.548875             20.655709         12.755661   
sheffield             4.631778              5.632625          1.706479   
simia                 8.312692              9.489426          5.088971   
soton                10.074514             11.831864          2.565766   
swcompany             0.000000              5.733333          0.000000   
tomheath             11.250000             15.750000          1.333333   
twrpi                 5.800000              8.636364          1.800000   
ukpostcodes           4.200000              9.400000          0.600000   

               econnprimecardinality  
dataset                               
aemet                       1.864662  
bfs                        26.091886  
bibbase                     7.359940  
bibsonomy                   3.012178  
bne                         8.396296  
dnb                        10.882485  
dwsmannheim                 7.577465  
eurostat                   16.699387  
eye48                       1.000000  
fao                         6.333333  
geovocab                    5.853546  
govwild                     0.000000  
harth                      12.586207  
icane                      12.409091  
imf                        24.032258  
korrekt                     9.373848  
l3s                         6.196454  
linkedgeodata              52.866545  
ndljp                       3.000000  
ontologi                  189.769231  
openei                     26.000000  
reegle                     82.979424  
revyu                       4.020900  
rodeionet                  34.458466  
semanticweb                13.865584  
sheffield                   2.707605  
simia                       6.214130  
soton                       4.425695  
swcompany                   5.033333  
tomheath                    6.166667  
twrpi                       3.927273  
ukpostcodes                 3.600000  
m12a
dataset
aemet             0.150376
bfs               4.709296
bibbase           0.796687
bibsonomy         0.002105
bne               0.000000
dnb               0.792945
dwsmannheim       1.028169
eurostat          0.733129
eye48             0.000000
fao               0.400000
geovocab          0.000000
govwild           0.000000
harth             1.296552
icane            29.613636
imf               7.568548
korrekt           0.000439
l3s               0.000362
linkedgeodata     3.020110
ndljp             0.000000
ontologi          0.221060
openei            0.000000
reegle           42.913580
revyu             0.506801
rodeionet         0.000000
semanticweb       0.099822
sheffield         0.000049
simia             0.098527
soton             0.132367
swcompany         0.000000
tomheath          0.083333
twrpi             0.218182
ukpostcodes       0.600000
Name: diffratio, dtype: float64
m12b
dataset
aemet            0.428552
bfs              1.225874
bibbase          0.615683
bibsonomy        0.003380
bne              0.282658
dnb              0.886173
dwsmannheim      0.271333
eurostat         0.738727
eye48            1.389975
fao              0.531327
geovocab         1.423988
govwild          0.000000
harth            1.830508
icane            1.619802
imf              1.364628
korrekt          0.230051
l3s              0.270947
linkedgeodata    1.642036
ndljp            0.000000
ontologi         0.302997
openei           0.000000
reegle           2.795777
revyu            0.610970
rodeionet        0.049154
semanticweb      0.284479
sheffield        0.293578
simia            0.265544
soton            0.336969
swcompany        0.714283
tomheath         0.324859
twrpi            0.379368
ukpostcodes      0.605829
Name: diffentropy, dtype: float64
m12c
               desccardinality  descprimecardinality
dataset                                             
aemet                 3.165414              4.142857
bfs                   3.147770              4.149919
bibbase               5.659639              6.243976
bibsonomy            19.737716             19.737744
bne                   3.592593              4.581481
dnb                   6.265925              7.246826
dwsmannheim           3.845070              4.845070
eurostat             12.623722             13.621677
eye48                 0.000000              0.333333
fao                   4.800000              5.866667
geovocab              0.000000              0.999878
govwild               0.000000              0.000000
harth                 2.620690              3.731034
icane                 1.863636              3.022727
imf                   3.129032              4.129032
korrekt              10.068341             11.067573
l3s                   7.526411              8.869754
linkedgeodata        46.478976             47.453382
ndljp                 8.000000              9.000000
ontologi             10.212845             12.707991
openei               18.750000             19.750000
reegle                5.144033              6.144033
revyu                 2.446450              3.441660
rodeionet            20.514377             21.734824
semanticweb          10.081175             11.086688
sheffield             3.926360              4.926077
simia                 8.155208              9.155552
soton                 5.154257              6.587740
swcompany             0.000000              1.000000
tomheath             10.250000             11.583333
twrpi                 4.872727              5.818182
ukpostcodes           4.200000              5.200000
m13a
dataset
aemet            0.075188
bfs              0.000000
bibbase          0.000000
bibsonomy        0.000000
bne              0.000000
dnb              0.000000
dwsmannheim      0.000000
eurostat         0.000000
eye48            0.000000
fao              0.000000
geovocab         0.000000
govwild          0.000000
harth            0.358621
icane            0.818182
imf              0.000000
korrekt          0.240237
l3s              0.000362
linkedgeodata    0.005484
ndljp            0.000000
ontologi         0.000000
openei           0.000000
reegle           0.000000
revyu            0.354196
rodeionet        0.000000
semanticweb      0.254199
sheffield        0.000049
simia            0.131300
soton            0.097655
swcompany        0.000000
tomheath         0.000000
twrpi            0.000000
ukpostcodes      0.000000
Name: diffratio, dtype: float64
m13b
dataset
aemet            0.422333
bfs              0.411396
bibbase          0.192189
bibsonomy        0.018212
bne              0.362053
dnb              0.223080
dwsmannheim      0.379662
eurostat         0.127210
eye48            0.000000
fao              0.323568
geovocab         0.000000
govwild          0.000000
harth            0.381323
icane            0.783015
imf              0.327240
korrekt          0.334833
l3s              0.296335
linkedgeodata    0.072744
ndljp            0.173965
ontologi         0.143763
openei           0.107676
reegle           0.330872
revyu            0.612852
rodeionet        0.113062
semanticweb      0.344873
sheffield        0.332856
simia            0.287971
soton            0.377083
swcompany        0.000000
tomheath         0.109035
twrpi            0.263884
ukpostcodes      0.134201
Name: diffentropy, dtype: float64
m13c
m21a
dataset
aemet             0.586466
bfs               7.312198
bibbase           0.000000
bibsonomy         0.019278
bne               0.151852
dnb               1.396360
dwsmannheim       0.887324
eurostat          5.805726
eye48             0.000000
fao               0.133333
geovocab          0.000000
govwild           0.000000
harth             0.193103
icane             0.250000
imf              10.310484
korrekt           0.000000
l3s               0.310058
linkedgeodata    23.071298
ndljp             0.000000
ontologi          1.112771
openei            0.000000
reegle           64.670782
revyu             0.225586
rodeionet         0.000000
semanticweb       0.002812
sheffield         0.574136
simia             0.055991
soton             0.109003
swcompany         0.000000
tomheath          1.416667
twrpi             1.000000
ukpostcodes       3.000000
Name: diffratio, dtype: float64
m21b
dataset
aemet            0.671593
bfs              1.411830
bibbase          0.644791
bibsonomy        0.592803
bne              0.504519
dnb              1.139924
dwsmannheim      0.764771
eurostat         1.688943
eye48            0.500000
fao              0.647883
geovocab         1.352330
govwild          0.000000
harth            1.598919
icane            1.294421
imf              1.507163
korrekt          0.201050
l3s              0.677691
linkedgeodata    2.907549
ndljp            0.584963
ontologi         0.657085
openei           0.047190
reegle           3.310206
revyu            0.353690
rodeionet        0.081979
semanticweb      0.216888
sheffield        0.784588
simia            0.262041
soton            0.268477
swcompany        0.947492
tomheath         0.798445
twrpi            0.636568
ukpostcodes      1.323526
Name: diffentropy, dtype: float64
m21c
               datasetcardinality  econnprimecardinality
dataset                                                 
aemet                    0.601504               1.669173
bfs                      0.486298               2.593767
bibbase                  0.367470               1.849398
bibsonomy                0.999972               1.999972
bne                      0.988889               2.022222
dnb                      0.998989               2.351534
dwsmannheim              1.000000               2.000000
eurostat                 0.997955               3.162577
eye48                    0.000000               0.333333
fao                      0.400000               2.066667
geovocab                 0.000000               2.032721
govwild                  0.000000               0.000000
harth                    0.186207               4.075862
icane                    0.272727               1.954545
imf                      0.483871               2.991935
korrekt                  0.759434               1.759105
l3s                      0.999638               2.998915
linkedgeodata            0.939671               2.579525
ndljp                    1.000000               2.000000
ontologi                 0.997760               2.823749
openei                   1.000000               2.000000
reegle                   0.851852               2.473251
revyu                    0.672330               1.687921
rodeionet                0.995208               2.103834
semanticweb              0.745782               1.822924
sheffield                0.996838               1.996887
simia                    0.862594               1.865664
soton                    0.410814               1.771285
swcompany                0.000000               1.366667
tomheath                 0.500000               3.333333
twrpi                    0.709091               1.963636
ukpostcodes              0.600000               1.400000
m22a
dataset
aemet            0.616541
bfs              1.082214
bibbase          0.367470
bibsonomy        0.999972
bne              1.033333
dnb              1.351534
dwsmannheim      1.000000
eurostat         2.164622
eye48            0.000000
fao              0.866667
geovocab         0.000000
govwild          0.000000
harth            1.558621
icane            0.340909
imf              1.491935
korrekt          0.759434
l3s              1.999276
linkedgeodata    1.570384
ndljp            1.000000
ontologi         1.825990
openei           1.000000
reegle           1.411523
revyu            0.687943
rodeionet        1.108626
semanticweb      0.822943
sheffield        0.996890
simia            0.865698
soton            0.770061
swcompany        0.000000
tomheath         2.416667
twrpi            0.963636
ukpostcodes      0.800000
Name: diffratio, dtype: float64
m22b
dataset
aemet            0.682342
bfs              1.108201
bibbase          0.467401
bibsonomy        0.918943
bne              0.785654
dnb              0.922768
dwsmannheim      0.870714
eurostat         1.451795
eye48            0.000000
fao              0.841343
geovocab         0.848196
govwild          0.000000
harth            1.039577
icane            0.762736
imf              1.237756
korrekt          0.413288
l3s              1.045159
linkedgeodata    1.214515
ndljp            0.918296
ontologi         0.573142
openei           0.306762
reegle           1.034891
revyu            0.520357
rodeionet        0.300281
semanticweb      0.447318
sheffield        0.936320
simia            0.556909
soton            0.410044
swcompany        0.301462
tomheath         0.990731
twrpi            0.792236
ukpostcodes      0.612197
Name: diffentropy, dtype: float64
m22c
               vocabdesccardinality  vocabdescprimecardinality
dataset                                                       
aemet                      1.548872                   1.977444
bfs                        2.008598                   2.861365
bibbase                    1.990964                   2.257530
bibsonomy                 14.195499                  14.195527
bne                        1.314815                   2.303704
dnb                        1.447028                   2.253342
dwsmannheim                2.169014                   3.169014
eurostat                   3.984663                   4.981595
eye48                      0.000000                   0.333333
fao                        2.266667                   2.933333
geovocab                   0.000000                   0.999878
govwild                    0.000000                   0.000000
harth                      0.820690                   1.600000
icane                      1.181818                   1.659091
imf                        2.008065                   2.846774
korrekt                    2.316257                   3.307372
l3s                        4.360347                   4.960564
linkedgeodata              4.733090                   5.707495
ndljp                      5.000000                   5.000000
ontologi                   3.658701                   4.401792
openei                     3.000000                   4.000000
reegle                     2.666667                   3.514403
revyu                      1.842514                   2.805052
rodeionet                  2.980831                   2.985623
semanticweb                2.649179                   3.552390
sheffield                  1.996821                   2.996537
simia                      2.643840                   3.642633
soton                      2.421884                   3.580564
swcompany                  0.000000                   1.000000
tomheath                   1.916667                   2.583333
twrpi                      1.872727                   2.727273
ukpostcodes                1.800000                   2.400000
m31a
dataset
aemet            0.0
bfs              0.0
bibbase          0.0
bibsonomy        0.0
bne              0.0
dnb              0.0
dwsmannheim      0.0
eurostat         0.0
eye48            0.0
fao              0.0
geovocab         0.0
govwild          0.0
harth            0.0
icane            0.0
imf              0.0
korrekt          0.0
l3s              0.0
linkedgeodata    0.0
ndljp            0.0
ontologi         0.0
openei           0.0
reegle           0.0
revyu            0.0
rodeionet        0.0
semanticweb      0.0
sheffield        0.0
simia            0.0
soton            0.0
swcompany        0.0
tomheath         0.0
twrpi            0.0
ukpostcodes      0.0
Name: diffratio, dtype: float64
m31b
dataset
aemet            0.259129
bfs              0.476778
bibbase          0.139191
bibsonomy        0.000000
bne              0.714848
dnb              0.443146
dwsmannheim      0.548015
eurostat         0.253429
eye48            0.000000
fao              0.341035
geovocab         0.000000
govwild          0.000000
harth            0.243550
icane            0.421112
imf              0.469375
korrekt          0.492164
l3s              0.229150
linkedgeodata    0.169645
ndljp            0.041521
ontologi         0.216640
openei           0.224535
reegle           0.360092
revyu            0.671730
rodeionet       -0.027343
semanticweb      0.444325
sheffield        0.561856
simia            0.444835
soton            0.636998
swcompany        0.000000
tomheath         0.284164
twrpi            0.506050
ukpostcodes      0.242018
Name: diffentropy, dtype: float64
m31c

correlation between ratio and diffentropy: do they really measure something different?


In [864]:
#def correlation_ratio_diffentropy(list):
     #concat = pd.concat(list, axis=1)
     #corr_bc = concat[['diffratio', 'diffentropy']]
         #print(corr_bc.corr(method='spearman')) #pearson normality
        #print(corr_bc.hist(column='diffentropy'))
        #print(concat.head
def correlation_ratio_difference(b,c):
    ratioandentropy=[]
    ratioandentropy.append(b['diffratio'])
    ratioandentropy.append(c['diffentropy'])
    bc = ratioandentropyinonetable(ratioandentropy)
    bc.head(2)
    print(b.head())
    print(c.head())
    print(bc.corr(method='spearman')) #pearson normality
    #spearmanr(b,c)
    #print(bc.hist(column='diffentropy'))

In [865]:
print(concat_m22bo.head())
print(str(concat_m11bi.count(axis=0)))


                                              entity typeentity  diffratio  \
0  http://dblp.l3s.de/d2r/resource/authors/Andrea...       null        1.0   
1  http://dblp.l3s.de/d2r/resource/authors/Susie_...       null        1.0   
2  http://dblp.l3s.de/d2r/resource/authors/Andrew...       null        1.0   
3  http://dblp.l3s.de/d2r/resource/publications/c...       null        2.0   
4  http://dblp.l3s.de/d2r/resource/publications/c...       null        2.0   

  dataset  
0     l3s  
1     l3s  
2     l3s  
3     l3s  
4     l3s  
entity        52774
typeentity    52780
diffratio     52780
dataset       52780
dtype: int64

In [866]:
print(concat_m22co.head())


                                              entity typeentity  diffentropy  \
0  http://dblp.l3s.de/d2r/resource/authors/Andrea...       null     1.000000   
1  http://dblp.l3s.de/d2r/resource/authors/Susie_...       null     1.000000   
2  http://dblp.l3s.de/d2r/resource/authors/Andrew...       null     1.000000   
3  http://dblp.l3s.de/d2r/resource/publications/c...       null     1.061278   
4  http://dblp.l3s.de/d2r/resource/publications/c...       null     0.699843   

  dataset  
0     l3s  
1     l3s  
2     l3s  
3     l3s  
4     l3s  

In [867]:
print(concat_m11ci.describe())

#concat_m11bi.isnull().values.any()


       diffentropy
count      52780.0
mean           0.0
std            0.0
min            0.0
25%            0.0
50%            0.0
75%            0.0
max            0.0

In [ ]:


In [868]:
if not concat_m11bi.empty and not concat_m11ci.empty:  # or for all rows (including NaN and null) len(concat_m11bi) !=0 and len(concat_m11ci) !=0
    correlation_ratio_diffentropy([concat_m11bi,concat_m11ci])
if not concat_m12bi.empty and not concat_m12ci.empty:     
    correlation_ratio_diffentropy([concat_m12bi,concat_m12ci])
if not concat_m13bi.empty and not concat_m13ci.empty: 
    correlation_ratio_diffentropy([concat_m13bi,concat_m13ci])
if not concat_m21bi.empty and not concat_m21ci.empty: 
    correlation_ratio_diffentropy([concat_m21bi,concat_m21ci])
if not concat_m22bi.empty and not concat_m22ci.empty: 
    correlation_ratio_diffentropy([concat_m22bi,concat_m22ci])
if not concat_m31bi.empty and not concat_m31ci.empty: 
    correlation_ratio_diffentropy([concat_m31bi,concat_m31ci])
    
#cor(i,j) = cov(i,j)/[stdev(i)*stdev(j)] will be Nan if denominator is 0


             diffratio  diffentropy
diffratio          NaN          NaN
diffentropy        NaN          NaN
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4afbcea908>]]
             diffratio  diffentropy
diffratio     1.000000     0.712856
diffentropy   0.712856     1.000000
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4aa7b37470>]]
             diffratio  diffentropy
diffratio     1.000000     0.015079
diffentropy   0.015079     1.000000
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4af1b8a048>]]
             diffratio  diffentropy
diffratio     1.000000     0.598509
diffentropy   0.598509     1.000000
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4ad5515358>]]
             diffratio  diffentropy
diffratio     1.000000     0.393945
diffentropy   0.393945     1.000000
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4af750ff98>]]
             diffratio  diffentropy
diffratio          NaN          NaN
diffentropy        NaN          1.0
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4aa7f32e48>]]

In [871]:
if not concat_m11bs.empty and not concat_m11cs.empty:
    correlation_ratio_diffentropy([concat_m11bs,concat_m11cs])
if not concat_m12bs.empty and not concat_m12cs.empty:
    correlation_ratio_diffentropy([concat_m12bs,concat_m12cs])
if not concat_m13bs.empty and not concat_m13cs.empty:
    correlation_ratio_diffentropy([concat_m13bs,concat_m13cs])
if not concat_m21bs.empty and not concat_m21cs.empty:
    correlation_ratio_diffentropy([concat_m21bs,concat_m21cs])
if not concat_m22bs.empty and not concat_m22cs.empty:
    correlation_ratio_diffentropy([concat_m22bs,concat_m22cs])
if not concat_m31bs.empty and not concat_m31cs.empty:
    correlation_ratio_diffentropy([concat_m31bs,concat_m31cs])


             diffratio  diffentropy
diffratio          NaN          NaN
diffentropy        NaN          NaN
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4aa77f2550>]]
             diffratio  diffentropy
diffratio          NaN          NaN
diffentropy        NaN          NaN
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4a620cc828>]]
             diffratio  diffentropy
diffratio          NaN          NaN
diffentropy        NaN          NaN
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4b1e1d4048>]]
             diffratio  diffentropy
diffratio          NaN          NaN
diffentropy        NaN          NaN
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4b462ac908>]]
             diffratio  diffentropy
diffratio          NaN          NaN
diffentropy        NaN          NaN
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4a60711630>]]
             diffratio  diffentropy
diffratio          NaN          NaN
diffentropy        NaN          NaN
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4aa77b28d0>]]

In [872]:
if not concat_m11br.empty and not concat_m11cr.empty:
    correlation_ratio_diffentropy([concat_m11br,concat_m11cr])
if not concat_m12br.empty and not concat_m12cr.empty:
    correlation_ratio_diffentropy([concat_m12br,concat_m12cr])
if not concat_m13br.empty and not concat_m13cr.empty:
    correlation_ratio_diffentropy([concat_m13br,concat_m13cr])
if not concat_m21br.empty and not concat_m21cr.empty:
    correlation_ratio_diffentropy([concat_m21br,concat_m21cr])
if not concat_m22br.empty and not concat_m22cr.empty:
    correlation_ratio_diffentropy([concat_m22br,concat_m22cr])
if not concat_m31br.empty and not concat_m31cr.empty:
    correlation_ratio_diffentropy([concat_m31br,concat_m31cr])


             diffratio  diffentropy
diffratio          NaN          NaN
diffentropy        NaN          NaN
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4b163a67f0>]]
             diffratio  diffentropy
diffratio      1.00000      0.60706
diffentropy    0.60706      1.00000
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4aa73b2080>]]
             diffratio  diffentropy
diffratio     1.000000     0.837708
diffentropy   0.837708     1.000000
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4aa706a8d0>]]
             diffratio  diffentropy
diffratio     1.000000     0.614191
diffentropy   0.614191     1.000000
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4afb619e48>]]
             diffratio  diffentropy
diffratio     1.000000     0.954121
diffentropy   0.954121     1.000000
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4af517ceb8>]]
             diffratio  diffentropy
diffratio          NaN          NaN
diffentropy        NaN          1.0
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4af52f3860>]]

In [873]:
if not concat_m11bo.empty and not concat_m11co.empty:
    correlation_ratio_diffentropy([concat_m11bo,concat_m11co])
if not concat_m12bo.empty and not concat_m12co.empty:
    correlation_ratio_diffentropy([concat_m12bo,concat_m12co])
if not concat_m13bo.empty and not concat_m13co.empty:
    correlation_ratio_diffentropy([concat_m13bo,concat_m13co])
if not concat_m21bo.empty and not concat_m21co.empty:
    correlation_ratio_diffentropy([concat_m21bo,concat_m21co])
if not concat_m22bo.empty and not concat_m22co.empty:
    correlation_ratio_diffentropy([concat_m22bo,concat_m22co])
if not concat_m31bo.empty and not concat_m31co.empty:
    correlation_ratio_diffentropy([concat_m31bo,concat_m31co])


             diffratio  diffentropy
diffratio          NaN          NaN
diffentropy        NaN          NaN
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4a7871cc18>]]
             diffratio  diffentropy
diffratio     1.000000     0.391484
diffentropy   0.391484     1.000000
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4af3bd0f98>]]
             diffratio  diffentropy
diffratio      1.00000      0.04478
diffentropy    0.04478      1.00000
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4aa768c9b0>]]
             diffratio  diffentropy
diffratio     1.000000     0.675234
diffentropy   0.675234     1.000000
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4a60596978>]]
             diffratio  diffentropy
diffratio     1.000000     0.219982
diffentropy   0.219982     1.000000
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4a701992b0>]]
             diffratio  diffentropy
diffratio          NaN          NaN
diffentropy        NaN          1.0
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4a5faf0588>]]

In [874]:
if not concat_m11bc.empty and not concat_m11cc.empty:
    correlation_ratio_diffentropy([concat_m11bc,concat_m11cc])
if not concat_m12bc.empty and not concat_m12cc.empty:
    correlation_ratio_diffentropy([concat_m12bc,concat_m12cc])
if not concat_m13bc.empty and not concat_m13cc.empty:
    correlation_ratio_diffentropy([concat_m13bc,concat_m13cc])
if not concat_m21bc.empty and not concat_m21cc.empty:
    correlation_ratio_diffentropy([concat_m21bc,concat_m21cc])
if not concat_m22bc.empty and not concat_m22cc.empty:
    correlation_ratio_diffentropy([concat_m22bc,concat_m22cc])
if not concat_m31bc.empty and not concat_m31cc.empty:
    correlation_ratio_diffentropy([concat_m31bc,concat_m31cc])


             diffratio  diffentropy
diffratio          NaN          NaN
diffentropy        NaN          NaN
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4a603d94a8>]]
             diffratio  diffentropy
diffratio     1.000000     0.327775
diffentropy   0.327775     1.000000
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4aeeabe908>]]
             diffratio  diffentropy
diffratio     1.000000     0.353579
diffentropy   0.353579     1.000000
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4aedc99860>]]
             diffratio  diffentropy
diffratio     1.000000     0.910328
diffentropy   0.910328     1.000000
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4af6ec7f98>]]
             diffratio  diffentropy
diffratio      1.00000      0.38765
diffentropy    0.38765      1.00000
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4a6012db38>]]
             diffratio  diffentropy
diffratio          NaN          NaN
diffentropy        NaN          1.0
[[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4b05defe10>]]

In [902]:
def boxplots_for_alldatasets_entropy(list):
    df = list[0]
    measure = list[1]
    i= list[2]
    print(measure)
    df.boxplot(column='diffentropy',by='dataset',rot=15,vert=False)
    #date =pd.to_datetime('today')
    plt.savefig('boxplot_'+measure+str(i)+'.png')

In [903]:
# only for the measures with entropy
print('m11c for all each type of link in all data sets ------------------------------------------')
boxplots_for_alldatasets_entropy([concat_m11ci,'m11c',1])
boxplots_for_alldatasets_entropy([concat_m11cs,'m11c',2])
boxplots_for_alldatasets_entropy([concat_m11cr,'m11c',3])
boxplots_for_alldatasets_entropy([concat_m11co,'m11c',4])
boxplots_for_alldatasets_entropy([concat_m11cc,'m11c',5])


print('m12c for all each type of link in all data sets ------------------------------------------')
boxplots_for_alldatasets_entropy([concat_m12ci,'m12c',1])
boxplots_for_alldatasets_entropy([concat_m12cs,'m12c',2])
boxplots_for_alldatasets_entropy([concat_m12cr,'m12c',3])
boxplots_for_alldatasets_entropy([concat_m12co,'m12c',4])
boxplots_for_alldatasets_entropy([concat_m12cc,'m12c',5])


print('m13c for all each type of link in all data sets ------------------------------------------')                               
boxplots_for_alldatasets_entropy([concat_m13ci,'m13c',1])
boxplots_for_alldatasets_entropy([concat_m13cs,'m13c',2])
boxplots_for_alldatasets_entropy([concat_m13cr,'m13c',3])
boxplots_for_alldatasets_entropy([concat_m13co,'m13c',4])
boxplots_for_alldatasets_entropy([concat_m13cc,'m13c',5])


print('m21c for all each type of link in all data sets ------------------------------------------')
boxplots_for_alldatasets_entropy([concat_m21ci,'m21c',1])
boxplots_for_alldatasets_entropy([concat_m21cs,'m21c',2])
boxplots_for_alldatasets_entropy([concat_m21cr,'m21c',3])
boxplots_for_alldatasets_entropy([concat_m21co,'m21c',4])
boxplots_for_alldatasets_entropy([concat_m21cc,'m21c',5])


print('m22c for all each type of link in all data sets ------------------------------------------')
boxplots_for_alldatasets_entropy([concat_m22ci,'m22c',1])
boxplots_for_alldatasets_entropy([concat_m22cs,'m22c',2])
boxplots_for_alldatasets_entropy([concat_m22cr,'m22c',3])
boxplots_for_alldatasets_entropy([concat_m22co,'m22c',4])
boxplots_for_alldatasets_entropy([concat_m22cc,'m22c',5])


print('m31c for all each type of link in all data sets ------------------------------------------')
boxplots_for_alldatasets_entropy([concat_m31ci,'m31c',1])
boxplots_for_alldatasets_entropy([concat_m31cs,'m31c',2])
boxplots_for_alldatasets_entropy([concat_m31cr,'m31c',3])
boxplots_for_alldatasets_entropy([concat_m31co,'m31c',4])
boxplots_for_alldatasets_entropy([concat_m31cc,'m31c',5])


m11c for all each type of link in all data sets ------------------------------------------
m11c
m11c
m11c
m11c
m11c
m12c for all each type of link in all data sets ------------------------------------------
m12c
m12c
m12c
m12c
m12c
m13c for all each type of link in all data sets ------------------------------------------
m13c
m13c
m13c
m13c
m13c
m21c for all each type of link in all data sets ------------------------------------------
m21c
m21c
m21c
m21c
m21c
m22c for all each type of link in all data sets ------------------------------------------
m22c
/home/alex/.virtualenvs/markov/lib/python3.5/site-packages/matplotlib/pyplot.py:524: RuntimeWarning: More than 20 figures have been opened. Figures created through the pyplot interface (`matplotlib.pyplot.figure`) are retained until explicitly closed and may consume too much memory. (To control this warning, see the rcParam `figure.max_open_warning`).
  max_open_warning, RuntimeWarning)
m22c
m22c
m22c
m22c
m31c for all each type of link in all data sets ------------------------------------------
m31c
m31c
m31c
m31c
m31c

In [877]:
linksdistributions = pd.read_csv('http://141.26.208.201/countlinks_09122016.csv',sep="\t",usecols=[0,1,2],dtype={'dataset':object, 'typelink':object, 'countlinks':float})
linksdistributions.head()


Out[877]:
dataset typelink countlinks
0 http://128.32.78.9/ c 3.0
1 http://128.32.78.9/ m 8.0
2 http://128.32.78.9/ r 17.0
3 http://128.32.78.9/ t 8.0
4 http://13mm.de c 11.0

In [878]:
linksdistributions.head()
#sum not hist
#linksdistributions.groupby(['dataset','typelink']).sum().sort().tail(10).plot.bar(y='countlinks')
#linksdistributions.groupby(['dataset','typelink']).hist(column='countlinks')


Out[878]:
dataset typelink countlinks
0 http://128.32.78.9/ c 3.0
1 http://128.32.78.9/ m 8.0
2 http://128.32.78.9/ r 17.0
3 http://128.32.78.9/ t 8.0
4 http://13mm.de c 11.0

correlation between measures


In [879]:
def correlation_measures(list):
    entropiesfrommeasures=[]
    df2=pd.DataFrame()
    i=0
    for df in list:      
        i=i+1   
        if i==1:
            df=df.rename(columns = {'diffentropy':'diffentropy_m11'})
        elif i==2:  
            df=df.rename(columns = {'diffentropy':'diffentropy_m12'})
        elif i==3:  
            df=df.rename(columns = {'diffentropy':'diffentropy_m13'})
        elif i==4:  
            df=df.rename(columns = {'diffentropy':'diffentropy_m21'})
        elif i==5:  
            df=df.rename(columns = {'diffentropy':'diffentropy_m22'})
        elif i==6:  
            df=df.rename(columns = {'diffentropy':'diffentropy_m31'})    
        #print(df.head(2))      
        entropiesfrommeasures.append(df)
        
    entropies= ratioandentropyinonetable(entropiesfrommeasures)
    #print(entropies.head(2))
   
    print(entropies.corr(method='spearman')) #pearson normality
    #spearmanr(b,c)

In [880]:
#concat_m11cc.head()
correlation_measures([concat_m11cc,concat_m12cc, concat_m13cc, concat_m21cc, concat_m22cc, concat_m31cc])


                 diffentropy_m11  diffentropy_m12  diffentropy_m13  \
diffentropy_m11              NaN              NaN              NaN   
diffentropy_m12              NaN         1.000000         0.116745   
diffentropy_m13              NaN         0.116745         1.000000   
diffentropy_m21              NaN         0.637627        -0.307347   
diffentropy_m22              NaN         0.629139        -0.321392   
diffentropy_m31              NaN         0.560923         0.740948   

                 diffentropy_m21  diffentropy_m22  diffentropy_m31  
diffentropy_m11              NaN              NaN              NaN  
diffentropy_m12         0.637627         0.629139         0.560923  
diffentropy_m13        -0.307347        -0.321392         0.740948  
diffentropy_m21         1.000000         0.980069         0.095156  
diffentropy_m22         0.980069         1.000000         0.070941  
diffentropy_m31         0.095156         0.070941         1.000000  

In [881]:
#########################################################################################################

In [882]:
m12csheffield.head()


Out[882]:
entity typeentity diffentropy diffNentropy

test function


In [883]:
def describeEntropies(df):
    # works for every row of the column passed when it's called with apply! e.g. m12csheffield.apply(describeEntropies,axis=1)
    print(df['diffentropy'].describe())
   
    #print(isinstance(df2, pd.Series))
    #df.describe(percentiles=[.05, .25, .75, .95])
    #df['diffNentropy'].describe(percentiles=[.05, .25, .75, .95])

In [884]:
describeEntropies(m12csheffield)
#m12csheffield.apply(describeEntropies,axis=1)
#m12csheffield['diffentropy'].describe()


count    0.0
mean     NaN
std      NaN
min      NaN
25%      NaN
50%      NaN
75%      NaN
max      NaN
Name: diffentropy, dtype: float64

In [885]:
m22bBIBBASE.hist(column='diffentropy',bins=20, log=True) #,by='typeentity'


Out[885]:
array([[<matplotlib.axes._subplots.AxesSubplot object at 0x7f4aed453780>]], dtype=object)

temp (was ok)


In [886]:
# add catalogue and function with for

#b 
m12bFile_l3s_r = pd.read_csv('http://141.26.208.201/m12bFile_l3s_r.tsv',sep="\t",usecols=[0,1,2],dtype={'entity':str, 'typeentity':object, 'diffratio':float})
m12bFile_bibbase_r = pd.read_csv('http://141.26.208.201/m12bFile_bibbase_r.tsv',sep="\t",usecols=[0,1,2],dtype={'entity':str, 'typeentity':object, 'diffratio':float})
m12bFile_sheffield_r = pd.read_csv('http://141.26.208.201/m12bFile_sheffield_r.tsv',sep="\t",usecols=[0,1,2],dtype={'entity':str, 'typeentity':object, 'diffratio':float})
m12bFile_bne_r = pd.read_csv('http://141.26.208.201/m12bFile_bne_r.tsv',sep="\t",usecols=[0,1,2],dtype={'entity':str, 'typeentity':object, 'diffratio':float})
m12bFile_soton_r = pd.read_csv('http://141.26.208.201/m12bFile_soton_r.tsv',sep="\t",usecols=[0,1,2],dtype={'entity':str, 'typeentity':object, 'diffratio':float})


## c
m12cFile_l3s_r = pd.read_csv('http://141.26.208.201/m12cFile_l3s_r.tsv',sep="\t",usecols=[0,1,2],dtype={'entity':str, 'typeentity':object, 'diffentropy':float})
m12cFile_bibbase_r = pd.read_csv('http://141.26.208.201/m12cFile_bibbase_r.tsv',sep="\t",usecols=[0,1,2],dtype={'entity':str, 'typeentity':object, 'diffentropy':float})
m12cFile_sheffield_r = pd.read_csv('http://141.26.208.201/m12cFile_sheffield_r.tsv',sep="\t",usecols=[0,1,2],dtype={'entity':str, 'typeentity':object, 'diffentropy':float})
m12cFile_bne_r = pd.read_csv('http://141.26.208.201/m12cFile_bne_r.tsv',sep="\t",usecols=[0,1,2],dtype={'entity':str, 'typeentity':object, 'diffentropy':float})
m12cFile_soton_r = pd.read_csv('http://141.26.208.201/m12cFile_soton_r.tsv',sep="\t",usecols=[0,1,2],dtype={'entity':str, 'typeentity':object, 'diffentropy':float})


---------------------------------------------------------------------------
HTTPError                                 Traceback (most recent call last)
<ipython-input-886-ba97284898b8> in <module>()
      2 
      3 #b
----> 4 m12bFile_l3s_r = pd.read_csv('http://141.26.208.201/m12bFile_l3s_r.tsv',sep="\t",usecols=[0,1,2],dtype={'entity':str, 'typeentity':object, 'diffratio':float})
      5 m12bFile_bibbase_r = pd.read_csv('http://141.26.208.201/m12bFile_bibbase_r.tsv',sep="\t",usecols=[0,1,2],dtype={'entity':str, 'typeentity':object, 'diffratio':float})
      6 m12bFile_sheffield_r = pd.read_csv('http://141.26.208.201/m12bFile_sheffield_r.tsv',sep="\t",usecols=[0,1,2],dtype={'entity':str, 'typeentity':object, 'diffratio':float})

/home/alex/.virtualenvs/markov/lib/python3.5/site-packages/pandas/io/parsers.py in parser_f(filepath_or_buffer, sep, delimiter, header, names, index_col, usecols, squeeze, prefix, mangle_dupe_cols, dtype, engine, converters, true_values, false_values, skipinitialspace, skiprows, nrows, na_values, keep_default_na, na_filter, verbose, skip_blank_lines, parse_dates, infer_datetime_format, keep_date_col, date_parser, dayfirst, iterator, chunksize, compression, thousands, decimal, lineterminator, quotechar, quoting, escapechar, comment, encoding, dialect, tupleize_cols, error_bad_lines, warn_bad_lines, skipfooter, skip_footer, doublequote, delim_whitespace, as_recarray, compact_ints, use_unsigned, low_memory, buffer_lines, memory_map, float_precision)
    643                     skip_blank_lines=skip_blank_lines)
    644 
--> 645         return _read(filepath_or_buffer, kwds)
    646 
    647     parser_f.__name__ = name

/home/alex/.virtualenvs/markov/lib/python3.5/site-packages/pandas/io/parsers.py in _read(filepath_or_buffer, kwds)
    372     filepath_or_buffer, _, compression = get_filepath_or_buffer(
    373         filepath_or_buffer, encoding,
--> 374         compression=kwds.get('compression', None))
    375     kwds['compression'] = (inferred_compression if compression == 'infer'
    376                            else compression)

/home/alex/.virtualenvs/markov/lib/python3.5/site-packages/pandas/io/common.py in get_filepath_or_buffer(filepath_or_buffer, encoding, compression)
    236 
    237     if _is_url(filepath_or_buffer):
--> 238         req = _urlopen(str(filepath_or_buffer))
    239         if compression == 'infer':
    240             content_encoding = req.headers.get('Content-Encoding', None)

/usr/lib/python3.5/urllib/request.py in urlopen(url, data, timeout, cafile, capath, cadefault, context)
    161     else:
    162         opener = _opener
--> 163     return opener.open(url, data, timeout)
    164 
    165 def install_opener(opener):

/usr/lib/python3.5/urllib/request.py in open(self, fullurl, data, timeout)
    470         for processor in self.process_response.get(protocol, []):
    471             meth = getattr(processor, meth_name)
--> 472             response = meth(req, response)
    473 
    474         return response

/usr/lib/python3.5/urllib/request.py in http_response(self, request, response)
    580         if not (200 <= code < 300):
    581             response = self.parent.error(
--> 582                 'http', request, response, code, msg, hdrs)
    583 
    584         return response

/usr/lib/python3.5/urllib/request.py in error(self, proto, *args)
    508         if http_err:
    509             args = (dict, 'default', 'http_error_default') + orig_args
--> 510             return self._call_chain(*args)
    511 
    512 # XXX probably also want an abstract factory that knows when it makes

/usr/lib/python3.5/urllib/request.py in _call_chain(self, chain, kind, meth_name, *args)
    442         for handler in handlers:
    443             func = getattr(handler, meth_name)
--> 444             result = func(*args)
    445             if result is not None:
    446                 return result

/usr/lib/python3.5/urllib/request.py in http_error_default(self, req, fp, code, msg, hdrs)
    588 class HTTPDefaultErrorHandler(BaseHandler):
    589     def http_error_default(self, req, fp, code, msg, hdrs):
--> 590         raise HTTPError(req.full_url, code, msg, hdrs, fp)
    591 
    592 class HTTPRedirectHandler(BaseHandler):

HTTPError: HTTP Error 404: Not Found

In [ ]:
concat_m12b =pd.concat([m12bFile_l3s_r,m12bFile_bibbase_r,m12bFile_sheffield_r,m12bFile_bne_r,m12bFile_soton_r])
#concat_m12b.head()
concat_m12c =pd.concat([m12cFile_l3s_r,m12cFile_bibbase_r,m12cFile_sheffield_r,m12cFile_bne_r,m12cFile_soton_r])
#concat_m12c.head()

correlation_ratio_diffentropy([concat_m12b,concat_m12c])

In [ ]:
#was wrong - all the same:
#correlation_ratio_diffentropy([m12bFile_l3s_i,m12bFile_bibbase_i,m12bFile_sheffield_i,m12bFile_bne_i,m12bFile_soton_i])

In [ ]:
datasets = ['reegle','revyu']

l2c_m11bi=[]
l2c_m12bi=[]
l2c_m13bi=[]
l2c_m21bi=[]
l2c_m22bi=[]
l2c_m31bi=[]

l2c_m11bs=[]
l2c_m12bs=[]
l2c_m13bs=[]
l2c_m21bs=[]
l2c_m22bs=[]
l2c_m31bs=[]

l2c_m11br=[]
l2c_m12br=[]
l2c_m13br=[]
l2c_m21br=[]
l2c_m22br=[]
l2c_m31br=[]

l2c_m11bo=[]
l2c_m12bo=[]
l2c_m13bo=[]
l2c_m21bo=[]
l2c_m22bo=[]
l2c_m31bo=[]

l2c_m11bc=[]
l2c_m12bc=[]
l2c_m13bc=[]
l2c_m21bc=[]
l2c_m22bc=[]
l2c_m31bc=[]

l2c_m11ci=[]
l2c_m12ci=[]
l2c_m13ci=[]
l2c_m21ci=[]
l2c_m22ci=[]
l2c_m31ci=[]

l2c_m11cs=[]
l2c_m12cs=[]
l2c_m13cs=[]
l2c_m21cs=[]
l2c_m22cs=[]
l2c_m31cs=[]

l2c_m11cr=[]
l2c_m12cr=[]
l2c_m13cr=[]
l2c_m21cr=[]
l2c_m22cr=[]
l2c_m31cr=[]

l2c_m11co=[]
l2c_m12co=[]
l2c_m13co=[]
l2c_m21co=[]
l2c_m22co=[]
l2c_m31co=[]

l2c_m11cc=[]
l2c_m12cc=[]
l2c_m13cc=[]
l2c_m21cc=[]
l2c_m22cc=[]
l2c_m31cc=[]


def getdata():
    for dataset in datasets:        
              #all measurements for links type='i'
        print('inside for of datasets')
        measurement=loadalldatasetsmeasurements(['m22','b',dataset,'r'])
        #print(dataset)
        #print(measurement)
        if len(measurement)==1:
            print('length == 1! the header plus one line'+dataset)
        elif not measurement.empty:
            l2c_m11bi.append(measurement)
            print('it is not empty'+dataset)                
        else:
            print('it is empty'+dataset)
            
        
getdata()

In [ ]:
print('len of l2c_m21br is'+str(len(l2c_m21br)))

print('len of l2c_m21cr is'+str(len(l2c_m21cr)))
if len(l2c_m21br) != 0 and len(l2c_m21cr)!= 0:
    concat_m21br =pd.concat(l2c_m21br)
    concat_m21cr = pd.concat(l2c_m21cr)
    inequalities(concat_m21br, concat_m21cr)

if len(l2c_m22br) != 0 and len(l2c_m22cr)!= 0:
    concat_m22br =pd.concat(l2c_m22br)
    concat_m22cr = pd.concat(l2c_m22cr)
    inequalities(concat_m22br, concat_m22cr)

In [ ]:
try:
    if concat_m11bi.empty and concat_m11ci.empty:
        print('m11 b/c i - empty')
    else:
        inequalities(concat_m11bi, concat_m11ci)

    if concat_m12bi.empty and concat_m12ci.empty:
        print('m12 b/c i - empty')
    else:    
        inequalities(concat_m12bi, concat_m12ci)

    if concat_m13bi.empty and concat_m13ci.empty:
        print('m13 b/c i - empty')
    else:
        inequalities(concat_m13bi, concat_m13ci)
    
    if concat_m21bi.empty and concat_m21ci.empty:
        print('m21 b/c i - empty') 
    else:    
        inequalities(concat_m21bi, concat_m21ci)
    
    if concat_m22bi.empty and concat_m22ci.empty:
        print('m22 b/c i - empty')  
    else:
        inequalities(concat_m22bi, concat_m22ci)

    if concat_m31bi.empty and concat_m31ci.empty:
        print('m31 b/c i - empty') 
    else:    
        inequalities(concat_m31bi, concat_m31ci) 
except NameError:
    print('caught NameError due to a data frame not having been created before because the list to concatenate was empty')

In [ ]:
try:
    if concat_m11bs.empty and concat_m11cs.empty:
        print('m11 b/c s - empty')
    else:
        inequalities(concat_m11bs, concat_m11cs)

    if concat_m12bs.empty and concat_m12cs.empty:
        print('m12 b/c s - empty')
    else:    
        inequalities(concat_m12bs, concat_m12cs)

    if concat_m13bs.empty and concat_m13cs.empty:
        print('m13 b/c s - empty')
    else:
        inequalities(concat_m13bs, concat_m13cs)
    
    if concat_m21bs.empty and concat_m21cs.empty:
        print('m21 b/c s - empty') 
    else:    
        inequalities(concat_m21bs, concat_m21cs)
    
    if concat_m22bs.empty and concat_m22cs.empty:
        print('m22 b/c s - empty')  
    else:
        inequalities(concat_m22bs, concat_m22cs)

    if concat_m31bs.empty and concat_m31cs.empty:
        print('m31 b/c s - empty') 
    else:    
        inequalities(concat_m31bs, concat_m31cs) 
except NameError:
    print('caught NameError due to a data frame not having been created before because the list to concatenate was empty')

In [ ]:
try:
    if concat_m11br.empty and concat_m11cr.empty:
        print('m11 b/c r - empty')
    else:
        inequalities(concat_m11br, concat_m11cr)

    if concat_m12br.empty and concat_m12cr.empty:
        print('m12 b/c r - empty')
    else:    
        inequalities(concat_m12br, concat_m12cr)

    if concat_m13br.empty and concat_m13cr.empty:
        print('m13 b/c r - empty')
    else:
        inequalities(concat_m13br, concat_m13cr)
    
    if concat_m21br.empty and concat_m21cr.empty:
        print('m21 b/c r - empty') 
    else:    
        inequalities(concat_m21br, concat_m21cr)
    
    if concat_m22br.empty and concat_m22cr.empty:
        print('m22 b/c r - empty')  
    else:
        inequalities(concat_m22br, concat_m22cr)

    if concat_m31br.empty and concat_m31cr.empty:
        print('m31 b/c r - empty') 
    else:    
        inequalities(concat_m31br, concat_m31cr) 
except NameError:
    print('caught NameError due to a data frame not having been created before because the list to concatenate was empty')

In [ ]:
try:
    if concat_m11bc.empty and concat_m11cc.empty:
        print('m11 b/c c - empty')
    else:
        inequalities(concat_m11bc, concat_m11cc)

    if concat_m12bc.empty and concat_m12cc.empty:
        print('m12 b/c c - empty')
    else:    
        inequalities(concat_m12bc, concat_m12cc)

    if concat_m13bc.empty and concat_m13cc.empty:
        print('m13 b/c c - empty')
    else:
        inequalities(concat_m13bc, concat_m13cc)
    
    if concat_m21bi.empty and concat_m21ci.empty:
        print('m21 b/c i - empty') 
    else:    
        inequalities(concat_m21bc, concat_m21cc)
    
    if concat_m22bc.empty and concat_m22cc.empty:
        print('m22 b/c c - empty')  
    else:
        inequalities(concat_m22bc, concat_m22cc)

    if concat_m31bc.empty and concat_m31cc.empty:
        print('m31 b/c c - empty') 
    else:    
        inequalities(concat_m31bc, concat_m31cc) 
except NameError:
    print('caught NameError due to a data frame not having been created before because the list to concatenate was empty')

In [ ]:
try:
    if concat_m11bo.empty and concat_m11co.empty:
        print('m11 b/c o - empty')
    else:
        inequalities(concat_m11bo, concat_m11co)

    if concat_m12bo.empty and concat_m12co.empty:
        print('m12 b/c o - empty')
    else:    
        inequalities(concat_m12bo, concat_m12co)

    if concat_m13bo.empty and concat_m13co.empty:
        print('m13 b/c o - empty')
    else:
        inequalities(concat_m13bo, concat_m13co)
    
    if concat_m21bo.empty and concat_m21co.empty:
        print('m21 b/c o - empty') 
    else:    
        inequalities(concat_m21bo, concat_m21co)
    
    if concat_m22bo.empty and concat_m22co.empty:
        print('m22 b/c o - empty')  
    else:
        inequalities(concat_m22bo, concat_m22co)

    if concat_m31bo.empty and concat_m31co.empty:
        print('m31 b/c o - empty') 
    else:    
        inequalities(concat_m31bo, concat_m31co) 
except NameError:
    print('caught NameError due to a data frame not having been created before because the list to concatenate was empty')