In [1]:
import pandas as pd
import py_stringmatching as sm
import math
import matplotlib.pyplot as plt
import numpy as np
from matplotlib import style
from sklearn.cross_validation import train_test_split
import re
import operator
style.use('ggplot')
# read in csv as dataframe
# c = pd.read_csv("../blocking/candidate_set.csv", encoding="ISO-8859-1", index_col='_id')
c = pd.read_csv("../blocking/candidate_set_with_added_features.csv", encoding="ISO-8859-1", index_col='_id')
#c = c.dropna()
c.head()
c:\users\aparn\anaconda\envs\my_first_env\lib\site-packages\sklearn\cross_validation.py:44: DeprecationWarning: This module was deprecated in version 0.18 in favor of the model_selection module into which all the refactored classes and functions are moved. Also note that the interface of the new CV iterators are different from that of this module. This module will be removed in 0.20.
"This module will be removed in 0.20.", DeprecationWarning)
Out[1]:
ltable_id
rtable_id
ltable_name
ltable_genus
ltable_family
ltable_ecology
ltable_countries
ltable_threat_paragraph
ltable_conservation_paragraph
ltable_pop_trend
...
countries_countries_lev_sim
countries_countries_nmw
countries_countries_sw
country_count_country_count_exm
country_count_country_count_anm
country_count_country_count_lev_dist
country_count_country_count_lev_sim
status_match
country_overlap
country_count_sim
_id
25215
1470
3
adelaide pigmy bluetongue skink, pygmy bluetongue
Tiliqua
Scincidae
NaN
Australia;
NaN
NaN
NaN
...
0.16
-10.0
3.0
0
0.5
1.0
0.0
1
0.0
1
75598
4363
9
albanian water frog
Pelophylax
Ranidae
Terrestrial; Freshwater
Albania; Montenegro;
\r\r\r\n The major threat is drainage of we...
["\n It is listed on Appendix III of the Be...
Decreasing
...
0.16
-2.0
10.0
1
1.0
0.0
1.0
1
1.0
1
75600
4365
9
NaN
Pelophylax
Ranidae
Terrestrial; Freshwater
China;
\r\r\r\n The major threat is over-collectin...
['\n The Shapotou location is within a nati...
Decreasing
...
0.08
-16.0
1.0
0
0.5
1.0
0.0
1
0.0
1
78778
7543
9
karpathos frog
Pelophylax
Ranidae
Terrestrial; Freshwater
Greece;
\r\r\r\n The major threat is habitat loss a...
['\n It is listed on Appendix III of the Be...
Decreasing
...
0.12
-14.0
2.0
0
0.5
1.0
0.0
0
0.0
1
83513
4363
10
albanian water frog
Pelophylax
Ranidae
Terrestrial; Freshwater
Albania; Montenegro;
\r\r\r\n The major threat is drainage of we...
["\n It is listed on Appendix III of the Be...
Decreasing
...
0.16
-2.0
10.0
1
1.0
0.0
1.0
1
1.0
1
5 rows × 50 columns
In [2]:
c.columns
Out[2]:
Index(['ltable_id', 'rtable_id', 'ltable_name', 'ltable_genus',
'ltable_family', 'ltable_ecology', 'ltable_countries',
'ltable_threat_paragraph', 'ltable_conservation_paragraph',
'ltable_pop_trend', 'ltable_status', 'ltable_country_count',
'ltable_scientific_name', 'rtable_scientific_name', 'rtable_name',
'rtable_kingdom', 'rtable_phylum', 'rtable_class', 'rtable_order',
'rtable_family', 'rtable_genus', 'rtable_size', 'rtable_threats',
'rtable_conservation', 'rtable_threat_keywords',
'rtable_conservation_keywords', 'rtable_status', 'rtable_countries',
'rtable_country_count', 'rtable_tCount', '_id.1', 'ltable_id.1',
'rtable_id.1', 'name_name_jac_qgm_3_qgm_3',
'name_name_jac_dlm_dc0_dlm_dc0', 'countries_countries_jac_qgm_3_qgm_3',
'countries_countries_cos_dlm_dc0_dlm_dc0',
'countries_countries_jac_dlm_dc0_dlm_dc0', 'countries_countries_mel',
'countries_countries_lev_dist', 'countries_countries_lev_sim',
'countries_countries_nmw', 'countries_countries_sw',
'country_count_country_count_exm', 'country_count_country_count_anm',
'country_count_country_count_lev_dist',
'country_count_country_count_lev_sim', 'status_match',
'country_overlap', 'country_count_sim'],
dtype='object')
In [3]:
def list_duplicates(seq):
seen = set()
seen_add = seen.add
# adds all elements it doesn't know yet to seen and all other to seen_twice
seen_twice = set( x for x in seq if x in seen or seen_add(x) )
# turn the set into a list (as requested)
return list( seen_twice )
list_duplicates(c.columns)
Out[3]:
[]
In [4]:
matches = 0
nonmatches = 0;
gold = None
DS = None
labels = []
while matches < 60:
print('matches:', matches)
gold,unlabeled = train_test_split(c, train_size=0.1551, test_size=0.8449)
#gold,unlabeled = train_test_split(c, train_size=0.486, test_size=0.514)
#Automatically label the training data
matches = 0;
nonmatches = 0;
labels = []
for index, row in gold.iterrows():
if row['rtable_scientific_name'].strip().lower() == row['ltable_scientific_name'].strip().lower():
labels.append(1)
matches = matches + 1
else:
labels.append(0)
nonmatches = nonmatches + 1
# matches = 50
gold['label'] = labels
#print('number of matches in training set:', matches)
#print('number of nonmatches in training set:', nonmatches)
matches: 0
matches: 30
matches: 45
matches: 43
matches: 37
matches: 37
matches: 34
matches: 38
matches: 41
matches: 44
matches: 41
matches: 42
matches: 40
matches: 46
matches: 50
matches: 47
matches: 36
matches: 49
matches: 46
matches: 39
matches: 39
matches: 46
matches: 50
matches: 38
matches: 30
matches: 39
matches: 34
matches: 44
matches: 46
matches: 40
matches: 48
matches: 41
matches: 40
matches: 39
matches: 40
matches: 46
matches: 32
matches: 39
matches: 47
matches: 35
matches: 39
matches: 39
matches: 37
matches: 43
matches: 32
matches: 34
matches: 46
matches: 33
matches: 36
matches: 38
matches: 42
matches: 39
matches: 39
matches: 49
matches: 40
matches: 37
matches: 36
matches: 35
matches: 30
matches: 30
matches: 51
matches: 28
matches: 44
matches: 43
matches: 28
matches: 40
matches: 35
matches: 49
matches: 42
matches: 39
matches: 41
matches: 46
matches: 34
matches: 45
matches: 45
matches: 41
matches: 33
matches: 46
matches: 32
matches: 41
matches: 26
matches: 30
matches: 37
matches: 38
matches: 34
matches: 38
matches: 43
matches: 41
matches: 39
matches: 43
matches: 40
matches: 43
matches: 40
matches: 38
matches: 31
matches: 29
matches: 41
matches: 35
matches: 33
matches: 43
matches: 41
matches: 46
matches: 44
matches: 51
matches: 41
matches: 47
matches: 42
matches: 25
matches: 45
matches: 32
matches: 35
matches: 33
matches: 46
matches: 37
matches: 39
matches: 42
matches: 35
matches: 39
matches: 43
matches: 32
matches: 36
matches: 31
matches: 35
matches: 55
matches: 38
matches: 43
matches: 41
matches: 47
matches: 42
matches: 43
matches: 34
matches: 34
matches: 36
matches: 41
matches: 43
matches: 46
matches: 44
matches: 43
matches: 42
matches: 42
matches: 31
matches: 43
matches: 34
matches: 39
matches: 50
matches: 38
matches: 49
matches: 43
matches: 37
matches: 35
matches: 43
matches: 32
matches: 41
matches: 40
matches: 32
matches: 42
matches: 40
matches: 32
matches: 39
matches: 40
matches: 43
matches: 55
matches: 34
matches: 44
matches: 35
matches: 39
matches: 53
matches: 36
matches: 44
matches: 28
matches: 36
matches: 41
matches: 36
matches: 48
matches: 53
matches: 38
matches: 49
matches: 37
matches: 43
matches: 40
matches: 52
matches: 39
matches: 38
matches: 45
matches: 38
matches: 40
matches: 35
matches: 37
matches: 40
matches: 49
matches: 44
matches: 45
matches: 36
matches: 42
matches: 35
matches: 46
matches: 42
matches: 37
matches: 39
matches: 38
matches: 45
matches: 42
matches: 46
matches: 34
matches: 47
matches: 51
matches: 37
matches: 39
matches: 32
matches: 40
matches: 41
matches: 52
matches: 39
matches: 44
matches: 49
matches: 28
matches: 54
matches: 34
matches: 32
matches: 33
matches: 50
matches: 43
matches: 45
matches: 36
matches: 42
matches: 49
matches: 40
matches: 33
matches: 44
matches: 47
matches: 34
matches: 41
matches: 52
matches: 42
matches: 43
matches: 44
matches: 37
matches: 38
matches: 43
matches: 40
matches: 39
matches: 38
matches: 39
matches: 38
matches: 47
matches: 35
matches: 39
matches: 40
matches: 42
matches: 35
matches: 44
matches: 35
matches: 39
matches: 58
matches: 32
matches: 34
matches: 45
matches: 38
matches: 39
matches: 38
matches: 38
matches: 44
matches: 42
matches: 32
matches: 28
matches: 48
matches: 51
matches: 45
matches: 38
matches: 35
matches: 37
matches: 33
matches: 37
matches: 44
matches: 49
matches: 30
matches: 37
matches: 43
matches: 40
matches: 34
matches: 39
matches: 37
matches: 36
matches: 43
matches: 44
matches: 39
matches: 31
matches: 37
matches: 32
matches: 43
matches: 40
matches: 32
matches: 44
matches: 44
matches: 33
matches: 37
matches: 38
matches: 48
matches: 39
matches: 40
matches: 39
matches: 47
matches: 39
matches: 48
matches: 41
matches: 43
matches: 42
matches: 47
matches: 45
matches: 36
matches: 41
matches: 41
matches: 51
matches: 29
matches: 33
matches: 44
matches: 33
matches: 41
matches: 36
matches: 52
matches: 32
matches: 43
matches: 41
matches: 43
matches: 41
matches: 35
matches: 44
matches: 34
matches: 45
matches: 42
matches: 46
matches: 38
matches: 41
matches: 44
matches: 49
matches: 45
matches: 44
matches: 31
matches: 40
matches: 42
matches: 42
matches: 52
matches: 41
matches: 38
matches: 44
matches: 40
matches: 41
matches: 46
matches: 35
matches: 37
matches: 41
matches: 40
matches: 38
matches: 44
matches: 35
matches: 30
matches: 33
matches: 44
matches: 37
matches: 42
matches: 36
matches: 43
matches: 32
matches: 45
matches: 39
matches: 39
matches: 45
matches: 38
matches: 42
matches: 39
matches: 42
matches: 42
matches: 52
matches: 39
matches: 40
matches: 32
matches: 49
matches: 34
matches: 39
matches: 36
matches: 34
matches: 36
matches: 42
matches: 39
matches: 36
matches: 43
matches: 35
matches: 47
matches: 43
matches: 42
matches: 42
matches: 48
matches: 42
matches: 37
matches: 43
matches: 39
matches: 40
matches: 46
matches: 40
matches: 44
matches: 35
matches: 37
matches: 38
matches: 43
matches: 34
matches: 50
matches: 45
matches: 34
matches: 26
matches: 39
matches: 41
matches: 42
matches: 39
matches: 34
matches: 44
matches: 40
matches: 44
matches: 39
matches: 48
matches: 47
matches: 40
matches: 44
matches: 33
matches: 47
matches: 45
matches: 37
matches: 48
matches: 40
matches: 35
matches: 40
matches: 28
matches: 44
matches: 38
matches: 35
matches: 42
matches: 42
matches: 42
matches: 38
matches: 48
matches: 38
matches: 46
matches: 51
matches: 42
matches: 43
matches: 37
matches: 42
matches: 42
matches: 31
matches: 33
matches: 40
matches: 39
matches: 41
matches: 44
matches: 41
matches: 41
matches: 44
matches: 47
matches: 32
matches: 42
matches: 38
matches: 45
matches: 37
matches: 45
matches: 38
matches: 41
matches: 43
matches: 42
matches: 36
matches: 31
matches: 48
matches: 48
matches: 42
matches: 48
matches: 31
matches: 37
matches: 42
matches: 36
matches: 37
matches: 43
matches: 40
matches: 44
matches: 36
matches: 38
matches: 31
matches: 28
matches: 59
matches: 38
matches: 40
matches: 40
matches: 47
matches: 40
matches: 35
matches: 43
matches: 42
matches: 40
matches: 32
matches: 39
matches: 41
matches: 44
matches: 51
matches: 31
matches: 44
matches: 33
matches: 44
matches: 28
matches: 39
matches: 44
matches: 34
matches: 33
matches: 45
matches: 39
matches: 39
matches: 35
matches: 24
matches: 35
matches: 37
matches: 38
matches: 39
matches: 40
matches: 43
matches: 42
matches: 49
matches: 42
matches: 34
matches: 41
matches: 44
matches: 34
matches: 36
matches: 33
matches: 48
matches: 40
matches: 51
matches: 38
matches: 37
matches: 47
matches: 45
matches: 41
matches: 48
matches: 33
matches: 45
matches: 42
matches: 37
matches: 54
matches: 48
matches: 45
matches: 38
matches: 38
matches: 33
matches: 41
matches: 36
matches: 39
matches: 34
matches: 45
matches: 33
matches: 37
matches: 36
matches: 38
matches: 42
matches: 50
matches: 47
matches: 38
matches: 47
matches: 39
matches: 41
matches: 42
matches: 44
matches: 37
matches: 49
matches: 33
matches: 32
matches: 44
matches: 48
matches: 35
matches: 33
matches: 35
matches: 41
matches: 49
matches: 37
matches: 33
matches: 44
matches: 45
matches: 40
matches: 43
matches: 32
matches: 45
matches: 35
matches: 41
matches: 40
matches: 31
matches: 51
matches: 45
matches: 46
matches: 40
matches: 35
matches: 52
matches: 40
matches: 54
matches: 33
matches: 46
matches: 34
matches: 41
matches: 34
matches: 40
matches: 35
matches: 35
matches: 49
matches: 37
matches: 48
matches: 47
matches: 49
matches: 36
matches: 34
matches: 41
matches: 47
matches: 46
matches: 48
matches: 42
matches: 42
matches: 49
matches: 41
matches: 39
matches: 36
matches: 34
matches: 36
matches: 54
matches: 46
matches: 33
matches: 35
matches: 35
matches: 38
matches: 41
matches: 32
matches: 33
matches: 43
matches: 46
matches: 47
matches: 48
matches: 30
matches: 44
matches: 36
matches: 40
matches: 39
matches: 42
matches: 43
matches: 29
matches: 37
matches: 34
matches: 36
matches: 31
matches: 47
matches: 42
matches: 42
matches: 47
matches: 40
matches: 39
matches: 31
matches: 33
matches: 34
matches: 32
matches: 30
matches: 48
matches: 37
matches: 32
matches: 33
matches: 39
matches: 34
matches: 33
matches: 36
matches: 46
matches: 38
matches: 40
matches: 42
matches: 39
matches: 42
matches: 43
matches: 40
matches: 36
matches: 41
matches: 34
matches: 40
matches: 45
matches: 37
matches: 30
matches: 33
matches: 45
matches: 31
matches: 36
matches: 43
matches: 41
matches: 36
matches: 43
matches: 37
matches: 34
matches: 42
matches: 45
matches: 48
matches: 40
matches: 47
matches: 47
matches: 36
matches: 41
matches: 36
matches: 33
matches: 42
matches: 36
matches: 39
matches: 39
matches: 39
matches: 31
matches: 40
matches: 45
matches: 38
matches: 37
matches: 43
matches: 42
matches: 36
matches: 54
matches: 43
matches: 35
matches: 41
matches: 35
matches: 45
matches: 38
matches: 40
matches: 37
matches: 41
matches: 33
matches: 41
matches: 43
matches: 38
matches: 40
matches: 42
matches: 36
matches: 45
matches: 38
matches: 39
matches: 45
matches: 40
matches: 45
matches: 41
matches: 35
matches: 44
matches: 41
matches: 34
matches: 32
matches: 39
matches: 40
matches: 48
matches: 38
matches: 36
matches: 38
matches: 42
matches: 33
matches: 31
matches: 42
matches: 38
matches: 33
matches: 36
matches: 48
matches: 40
matches: 41
matches: 37
matches: 37
matches: 45
matches: 33
matches: 40
matches: 47
matches: 38
matches: 37
matches: 46
matches: 48
matches: 38
matches: 35
matches: 51
matches: 42
matches: 47
matches: 40
matches: 42
matches: 46
matches: 31
matches: 47
matches: 39
matches: 40
matches: 41
matches: 44
matches: 48
matches: 51
matches: 44
matches: 34
matches: 44
matches: 49
matches: 48
matches: 45
matches: 43
matches: 52
matches: 44
matches: 39
matches: 43
matches: 44
matches: 45
matches: 31
matches: 49
matches: 36
matches: 33
matches: 42
matches: 30
matches: 34
matches: 42
matches: 36
matches: 39
matches: 41
matches: 48
matches: 41
matches: 40
matches: 33
matches: 46
matches: 38
matches: 40
matches: 39
matches: 38
matches: 55
matches: 39
matches: 39
matches: 50
matches: 36
matches: 35
matches: 33
matches: 45
matches: 41
matches: 31
matches: 36
matches: 33
matches: 38
matches: 44
matches: 36
matches: 46
matches: 39
matches: 42
matches: 42
matches: 33
matches: 43
matches: 34
matches: 43
matches: 34
matches: 35
matches: 37
matches: 40
matches: 39
matches: 41
matches: 32
matches: 36
matches: 30
matches: 38
matches: 39
matches: 37
matches: 40
matches: 44
matches: 46
matches: 51
matches: 39
matches: 43
matches: 48
matches: 45
matches: 44
matches: 43
matches: 44
matches: 42
matches: 27
matches: 39
matches: 46
matches: 42
matches: 35
matches: 46
matches: 44
matches: 42
matches: 39
matches: 36
matches: 42
matches: 43
matches: 39
matches: 49
matches: 34
matches: 33
matches: 46
matches: 37
matches: 43
matches: 34
matches: 42
matches: 42
matches: 39
matches: 35
matches: 32
matches: 27
matches: 36
matches: 46
matches: 37
matches: 45
matches: 38
matches: 41
matches: 41
matches: 37
matches: 30
matches: 38
matches: 43
matches: 39
matches: 45
matches: 39
matches: 46
matches: 43
matches: 47
matches: 40
matches: 38
matches: 34
matches: 32
matches: 36
matches: 35
matches: 38
matches: 29
matches: 36
matches: 45
matches: 47
matches: 47
matches: 33
matches: 37
matches: 46
matches: 47
matches: 36
matches: 37
matches: 43
matches: 35
matches: 52
matches: 37
matches: 39
matches: 44
matches: 41
matches: 34
matches: 36
matches: 41
matches: 41
matches: 37
matches: 42
matches: 36
matches: 38
matches: 37
matches: 35
matches: 40
matches: 35
matches: 36
matches: 43
matches: 43
matches: 48
matches: 45
matches: 36
matches: 39
matches: 30
matches: 44
matches: 45
matches: 30
matches: 37
matches: 37
matches: 45
matches: 37
matches: 48
matches: 47
matches: 49
matches: 33
matches: 40
matches: 32
matches: 42
matches: 45
matches: 37
matches: 35
matches: 44
matches: 35
matches: 37
matches: 32
matches: 33
matches: 34
matches: 40
matches: 34
matches: 39
matches: 36
matches: 31
matches: 34
matches: 42
matches: 48
matches: 31
matches: 31
matches: 36
matches: 40
matches: 40
matches: 41
matches: 44
matches: 49
matches: 37
matches: 48
matches: 30
matches: 41
matches: 42
matches: 40
matches: 44
matches: 46
matches: 42
matches: 38
matches: 33
matches: 28
matches: 41
matches: 40
matches: 41
matches: 32
matches: 48
matches: 38
matches: 30
matches: 49
matches: 49
matches: 42
matches: 44
matches: 40
matches: 37
matches: 41
matches: 28
matches: 39
matches: 39
matches: 44
matches: 40
matches: 35
matches: 41
matches: 38
matches: 38
matches: 38
matches: 42
matches: 31
matches: 38
matches: 39
matches: 32
matches: 32
matches: 39
matches: 46
matches: 41
matches: 47
matches: 46
matches: 42
matches: 45
matches: 41
matches: 42
matches: 45
matches: 55
matches: 51
matches: 39
matches: 43
matches: 34
matches: 39
matches: 34
matches: 33
matches: 38
matches: 34
matches: 42
matches: 43
matches: 37
matches: 37
matches: 35
matches: 42
matches: 51
matches: 41
matches: 33
matches: 31
matches: 42
matches: 41
matches: 40
matches: 37
matches: 36
matches: 39
matches: 36
matches: 43
matches: 48
matches: 39
matches: 36
matches: 46
matches: 37
matches: 33
matches: 43
matches: 34
matches: 43
matches: 42
matches: 29
matches: 42
matches: 44
matches: 41
matches: 38
matches: 34
matches: 44
matches: 36
matches: 52
matches: 44
matches: 39
matches: 37
matches: 44
matches: 34
matches: 41
matches: 39
matches: 48
matches: 41
matches: 35
matches: 43
matches: 38
matches: 30
matches: 34
matches: 36
matches: 36
matches: 48
matches: 39
matches: 48
matches: 46
matches: 46
matches: 31
matches: 35
matches: 46
matches: 36
matches: 44
matches: 35
matches: 40
matches: 39
matches: 34
matches: 51
matches: 39
matches: 40
matches: 50
matches: 35
matches: 42
matches: 32
matches: 35
matches: 40
matches: 35
matches: 37
matches: 36
matches: 40
matches: 44
matches: 28
matches: 43
matches: 53
matches: 40
matches: 32
matches: 30
matches: 49
matches: 35
matches: 38
matches: 41
matches: 42
matches: 51
matches: 45
matches: 45
matches: 41
matches: 48
matches: 47
matches: 34
matches: 38
matches: 39
matches: 36
matches: 39
matches: 52
matches: 40
matches: 36
matches: 43
matches: 43
matches: 44
matches: 37
matches: 40
matches: 37
matches: 37
matches: 30
matches: 39
matches: 42
matches: 39
matches: 30
matches: 38
matches: 31
matches: 41
matches: 41
matches: 45
matches: 34
matches: 48
matches: 42
matches: 47
matches: 30
matches: 38
matches: 43
matches: 46
matches: 53
matches: 44
matches: 38
matches: 44
matches: 49
matches: 33
matches: 50
matches: 42
matches: 35
matches: 44
matches: 35
matches: 40
matches: 31
matches: 47
matches: 39
matches: 43
matches: 45
matches: 44
matches: 32
matches: 40
matches: 23
matches: 42
matches: 47
matches: 36
matches: 37
matches: 39
matches: 38
matches: 36
matches: 43
matches: 39
matches: 41
matches: 37
matches: 36
matches: 36
matches: 34
matches: 46
matches: 40
matches: 46
matches: 52
matches: 41
matches: 48
matches: 30
matches: 40
matches: 37
matches: 45
matches: 36
matches: 38
matches: 28
matches: 35
matches: 44
matches: 38
matches: 41
matches: 51
matches: 34
matches: 43
matches: 33
matches: 47
matches: 31
matches: 34
matches: 33
matches: 38
matches: 40
matches: 42
matches: 32
matches: 38
matches: 37
matches: 37
matches: 42
matches: 30
matches: 31
matches: 45
matches: 43
matches: 47
matches: 40
matches: 41
matches: 39
matches: 46
matches: 38
matches: 48
matches: 38
matches: 41
matches: 31
matches: 43
matches: 39
matches: 43
matches: 30
matches: 36
matches: 51
matches: 44
matches: 45
matches: 35
matches: 43
matches: 37
matches: 44
matches: 40
matches: 39
matches: 46
matches: 38
matches: 34
matches: 37
matches: 41
matches: 43
matches: 50
matches: 36
matches: 42
matches: 37
matches: 39
matches: 47
matches: 33
matches: 32
matches: 37
matches: 40
matches: 34
matches: 43
matches: 43
matches: 34
matches: 32
matches: 37
matches: 42
matches: 45
matches: 43
matches: 46
matches: 35
matches: 38
matches: 41
matches: 39
matches: 40
matches: 41
matches: 37
matches: 39
matches: 39
matches: 41
matches: 39
matches: 36
matches: 45
matches: 47
matches: 35
matches: 45
matches: 34
matches: 36
matches: 40
matches: 31
matches: 48
matches: 46
matches: 40
matches: 41
matches: 42
matches: 41
matches: 42
matches: 35
matches: 41
matches: 47
matches: 50
matches: 42
matches: 40
matches: 39
matches: 38
matches: 35
matches: 38
matches: 34
matches: 31
matches: 27
matches: 40
matches: 42
matches: 47
matches: 35
matches: 39
matches: 45
matches: 33
matches: 36
matches: 43
matches: 33
matches: 26
matches: 27
matches: 40
matches: 39
matches: 40
matches: 53
matches: 46
matches: 40
matches: 32
matches: 51
matches: 55
matches: 26
matches: 35
matches: 44
matches: 43
matches: 43
matches: 37
matches: 35
matches: 46
matches: 39
matches: 44
matches: 47
matches: 45
matches: 35
matches: 38
matches: 38
matches: 40
matches: 38
matches: 40
matches: 52
matches: 42
matches: 48
matches: 39
matches: 35
matches: 34
matches: 43
matches: 44
matches: 47
matches: 34
matches: 41
matches: 40
matches: 31
matches: 42
matches: 36
matches: 44
matches: 32
matches: 37
matches: 53
matches: 36
matches: 50
matches: 33
matches: 42
matches: 37
matches: 35
matches: 30
matches: 36
matches: 35
matches: 45
matches: 42
matches: 43
matches: 36
matches: 39
matches: 47
matches: 36
matches: 32
matches: 38
matches: 40
matches: 48
matches: 51
matches: 44
matches: 46
matches: 40
matches: 38
matches: 36
matches: 44
matches: 34
matches: 35
matches: 40
matches: 34
matches: 37
matches: 37
matches: 43
matches: 37
matches: 37
matches: 50
matches: 36
matches: 35
matches: 38
matches: 36
matches: 39
matches: 44
matches: 41
matches: 35
matches: 53
matches: 40
matches: 32
matches: 52
matches: 39
matches: 37
matches: 41
matches: 41
matches: 49
matches: 35
matches: 40
matches: 44
matches: 34
matches: 43
matches: 42
matches: 31
matches: 37
matches: 33
matches: 35
matches: 39
matches: 35
matches: 46
matches: 38
matches: 28
matches: 37
matches: 46
matches: 40
matches: 44
matches: 37
matches: 36
matches: 34
matches: 43
matches: 44
matches: 40
matches: 40
matches: 34
matches: 38
matches: 41
matches: 41
matches: 29
matches: 38
matches: 37
matches: 40
matches: 35
matches: 37
matches: 41
matches: 44
matches: 47
matches: 43
matches: 36
matches: 36
matches: 40
matches: 41
matches: 28
matches: 47
matches: 35
matches: 43
matches: 36
matches: 44
matches: 29
matches: 33
matches: 37
matches: 44
matches: 41
matches: 42
matches: 40
matches: 38
matches: 39
matches: 46
matches: 30
matches: 33
matches: 41
matches: 42
matches: 40
matches: 33
matches: 43
matches: 36
matches: 43
matches: 27
matches: 50
matches: 46
matches: 43
matches: 32
matches: 37
matches: 41
matches: 37
matches: 38
matches: 42
matches: 47
matches: 48
matches: 42
matches: 39
matches: 38
matches: 47
matches: 37
matches: 37
matches: 46
matches: 43
matches: 39
matches: 34
matches: 42
matches: 35
matches: 37
matches: 30
matches: 43
matches: 38
matches: 27
matches: 44
matches: 43
matches: 43
matches: 46
matches: 34
matches: 40
matches: 44
matches: 36
matches: 37
matches: 41
matches: 46
matches: 47
matches: 34
matches: 35
matches: 35
matches: 36
matches: 31
matches: 42
matches: 38
matches: 41
matches: 46
matches: 46
matches: 32
matches: 38
matches: 56
matches: 41
matches: 38
matches: 36
matches: 41
matches: 41
matches: 54
matches: 44
matches: 35
matches: 46
matches: 46
matches: 46
matches: 44
matches: 39
matches: 42
matches: 42
matches: 39
matches: 36
matches: 39
matches: 50
matches: 38
matches: 32
matches: 48
matches: 29
matches: 47
matches: 40
matches: 38
matches: 31
matches: 39
matches: 42
matches: 45
matches: 42
matches: 35
matches: 33
matches: 31
matches: 37
matches: 25
matches: 46
matches: 44
matches: 40
matches: 36
matches: 43
matches: 52
matches: 39
matches: 46
matches: 33
matches: 48
matches: 43
matches: 39
matches: 46
matches: 43
matches: 32
matches: 31
matches: 38
matches: 35
matches: 38
matches: 46
matches: 45
matches: 33
matches: 37
matches: 38
matches: 39
matches: 36
matches: 45
matches: 42
matches: 46
matches: 44
matches: 46
matches: 47
matches: 42
matches: 36
matches: 52
matches: 30
matches: 43
matches: 40
matches: 44
matches: 35
matches: 39
matches: 40
matches: 43
matches: 38
matches: 30
matches: 37
matches: 33
matches: 41
matches: 35
matches: 32
matches: 36
matches: 45
matches: 40
matches: 36
matches: 41
matches: 39
matches: 38
matches: 35
matches: 44
matches: 44
matches: 41
matches: 53
matches: 40
matches: 47
matches: 46
matches: 44
matches: 40
matches: 36
matches: 40
matches: 31
matches: 35
matches: 44
matches: 47
matches: 39
matches: 42
matches: 47
matches: 39
matches: 45
matches: 41
matches: 40
matches: 45
matches: 30
matches: 42
matches: 45
matches: 38
matches: 37
matches: 52
matches: 44
matches: 37
matches: 42
matches: 38
matches: 42
matches: 43
matches: 30
matches: 48
matches: 46
matches: 35
matches: 40
matches: 42
matches: 37
matches: 35
matches: 39
matches: 37
matches: 39
matches: 41
matches: 38
matches: 38
matches: 35
matches: 37
matches: 47
matches: 39
matches: 38
matches: 43
matches: 36
matches: 45
matches: 41
matches: 38
matches: 44
matches: 40
matches: 41
matches: 39
matches: 42
matches: 45
matches: 48
matches: 42
matches: 42
matches: 35
matches: 36
matches: 46
matches: 34
matches: 42
matches: 36
matches: 42
matches: 42
matches: 35
matches: 43
matches: 43
matches: 28
matches: 36
matches: 38
matches: 36
matches: 36
matches: 42
matches: 44
matches: 43
matches: 44
matches: 55
matches: 49
matches: 39
matches: 42
matches: 46
matches: 44
matches: 38
matches: 38
matches: 40
matches: 30
matches: 43
matches: 46
matches: 40
matches: 43
matches: 41
matches: 27
matches: 42
matches: 38
matches: 39
matches: 40
matches: 40
matches: 54
matches: 47
matches: 36
matches: 36
matches: 43
matches: 32
matches: 45
matches: 41
matches: 36
matches: 38
matches: 40
matches: 43
matches: 41
matches: 39
matches: 33
matches: 47
matches: 36
matches: 44
matches: 41
matches: 44
matches: 40
matches: 41
matches: 42
matches: 35
matches: 32
matches: 48
matches: 41
matches: 27
matches: 40
matches: 28
matches: 36
matches: 47
matches: 40
matches: 36
matches: 35
matches: 31
matches: 40
matches: 32
matches: 38
matches: 39
matches: 48
matches: 38
matches: 34
matches: 29
matches: 28
matches: 35
matches: 28
matches: 32
matches: 40
matches: 42
matches: 38
matches: 39
matches: 47
matches: 41
matches: 37
matches: 36
matches: 40
matches: 36
matches: 51
matches: 39
matches: 44
matches: 36
matches: 33
matches: 34
matches: 26
matches: 34
matches: 42
matches: 41
matches: 48
matches: 38
matches: 32
matches: 48
matches: 35
matches: 23
matches: 34
matches: 41
matches: 45
matches: 39
matches: 37
matches: 44
matches: 46
matches: 46
matches: 43
matches: 37
matches: 43
matches: 36
matches: 33
matches: 39
matches: 38
matches: 42
matches: 39
matches: 37
matches: 41
matches: 35
matches: 39
matches: 36
matches: 37
matches: 32
matches: 37
matches: 50
matches: 49
matches: 46
matches: 41
matches: 33
matches: 41
matches: 45
matches: 36
matches: 34
matches: 44
matches: 33
matches: 41
matches: 48
matches: 38
matches: 44
matches: 34
matches: 47
matches: 32
matches: 38
matches: 37
matches: 39
matches: 43
matches: 32
matches: 39
matches: 44
matches: 43
matches: 38
matches: 30
matches: 36
matches: 46
matches: 48
matches: 36
matches: 38
matches: 34
matches: 46
matches: 47
matches: 38
matches: 42
matches: 34
matches: 39
matches: 38
matches: 32
matches: 38
matches: 35
matches: 48
matches: 30
matches: 37
matches: 37
matches: 44
matches: 47
matches: 47
matches: 42
matches: 45
matches: 37
matches: 40
matches: 58
matches: 40
matches: 38
matches: 29
matches: 50
matches: 32
matches: 42
matches: 35
matches: 37
matches: 28
matches: 32
matches: 36
matches: 39
matches: 33
matches: 34
matches: 39
matches: 40
matches: 45
matches: 44
matches: 37
matches: 36
matches: 40
matches: 29
matches: 37
matches: 42
matches: 41
matches: 36
matches: 46
matches: 46
matches: 40
matches: 49
matches: 42
matches: 35
matches: 48
matches: 40
matches: 54
matches: 35
matches: 41
matches: 40
matches: 44
matches: 44
matches: 46
matches: 43
matches: 47
matches: 40
matches: 33
matches: 39
matches: 46
matches: 43
matches: 37
matches: 45
matches: 50
matches: 39
matches: 33
matches: 46
matches: 33
matches: 36
matches: 42
matches: 30
matches: 44
matches: 30
matches: 42
matches: 39
matches: 37
matches: 43
matches: 36
matches: 40
matches: 43
matches: 39
matches: 41
matches: 34
matches: 38
matches: 35
matches: 44
matches: 42
matches: 48
matches: 49
matches: 37
matches: 39
matches: 51
matches: 42
matches: 36
matches: 34
matches: 33
matches: 53
matches: 28
matches: 45
matches: 52
matches: 52
matches: 35
matches: 41
matches: 45
matches: 36
matches: 42
matches: 41
matches: 32
matches: 46
matches: 51
matches: 37
matches: 45
matches: 39
matches: 48
matches: 38
matches: 49
matches: 42
matches: 41
matches: 41
matches: 43
matches: 36
matches: 43
matches: 38
matches: 42
matches: 36
matches: 38
matches: 36
matches: 46
matches: 35
matches: 47
matches: 46
matches: 32
matches: 47
matches: 40
matches: 47
matches: 33
matches: 41
matches: 41
matches: 35
matches: 34
matches: 39
matches: 37
matches: 34
matches: 29
matches: 36
matches: 47
matches: 38
matches: 39
matches: 37
matches: 41
matches: 44
matches: 42
matches: 33
matches: 35
matches: 42
matches: 44
matches: 37
matches: 43
matches: 48
matches: 42
matches: 41
matches: 38
matches: 44
matches: 41
matches: 30
matches: 42
matches: 45
matches: 43
matches: 43
matches: 38
matches: 36
matches: 44
matches: 40
matches: 32
matches: 29
matches: 35
matches: 38
matches: 46
matches: 35
matches: 34
matches: 38
matches: 45
matches: 39
matches: 38
matches: 39
matches: 30
matches: 45
matches: 37
matches: 43
matches: 43
matches: 42
matches: 37
matches: 52
matches: 36
matches: 50
matches: 52
matches: 40
matches: 43
matches: 53
matches: 37
matches: 44
matches: 29
matches: 43
matches: 41
matches: 41
matches: 37
matches: 38
matches: 42
matches: 37
matches: 52
matches: 40
matches: 36
matches: 44
matches: 34
matches: 36
matches: 42
matches: 40
matches: 45
matches: 46
matches: 40
matches: 35
matches: 35
matches: 41
matches: 31
matches: 35
matches: 35
matches: 40
matches: 37
matches: 34
matches: 42
matches: 49
matches: 29
matches: 51
matches: 31
matches: 41
matches: 40
matches: 37
matches: 45
matches: 41
matches: 39
matches: 39
matches: 41
matches: 36
matches: 39
matches: 39
matches: 41
matches: 40
matches: 42
matches: 34
matches: 38
matches: 46
matches: 38
matches: 39
matches: 42
matches: 46
matches: 43
matches: 34
matches: 42
matches: 36
matches: 42
matches: 45
matches: 44
matches: 32
matches: 48
matches: 46
matches: 33
matches: 35
matches: 34
matches: 34
matches: 34
matches: 39
matches: 40
matches: 36
matches: 43
matches: 41
matches: 42
matches: 49
matches: 48
matches: 44
matches: 35
matches: 39
matches: 37
matches: 34
matches: 21
matches: 42
matches: 41
matches: 33
matches: 46
matches: 37
matches: 39
matches: 41
matches: 40
matches: 35
matches: 39
matches: 43
matches: 34
matches: 41
matches: 45
matches: 46
matches: 40
matches: 40
matches: 35
matches: 44
matches: 43
matches: 39
matches: 37
matches: 36
matches: 43
matches: 39
matches: 43
matches: 42
matches: 39
matches: 30
matches: 42
matches: 39
matches: 41
matches: 42
matches: 48
matches: 39
matches: 42
matches: 44
matches: 32
matches: 53
matches: 37
matches: 47
matches: 45
matches: 45
matches: 45
matches: 48
matches: 38
matches: 48
matches: 48
matches: 36
matches: 39
matches: 46
matches: 47
matches: 50
matches: 43
matches: 42
matches: 44
matches: 36
matches: 42
matches: 34
matches: 36
matches: 39
matches: 44
matches: 38
matches: 35
matches: 44
matches: 29
matches: 34
matches: 45
matches: 42
matches: 52
matches: 27
matches: 40
matches: 33
matches: 47
matches: 50
matches: 35
matches: 37
matches: 40
matches: 29
matches: 33
matches: 26
matches: 45
matches: 38
matches: 44
matches: 39
matches: 36
matches: 41
matches: 41
matches: 36
matches: 41
matches: 33
matches: 44
matches: 35
matches: 56
matches: 44
matches: 40
matches: 36
matches: 40
matches: 42
matches: 32
matches: 30
matches: 45
matches: 38
matches: 39
matches: 39
matches: 39
matches: 38
matches: 38
matches: 46
matches: 34
matches: 44
matches: 42
matches: 36
matches: 39
matches: 42
matches: 26
matches: 39
matches: 42
matches: 46
matches: 32
matches: 41
matches: 43
matches: 35
matches: 43
matches: 49
matches: 53
matches: 36
matches: 40
matches: 36
matches: 31
matches: 45
matches: 45
matches: 37
matches: 45
matches: 37
matches: 35
matches: 35
matches: 39
matches: 37
matches: 37
matches: 37
matches: 40
matches: 40
matches: 41
matches: 41
matches: 40
matches: 46
matches: 44
matches: 35
matches: 48
matches: 31
matches: 47
matches: 39
matches: 40
matches: 38
matches: 38
matches: 41
matches: 34
matches: 33
matches: 43
matches: 41
matches: 41
matches: 46
matches: 36
matches: 31
matches: 37
matches: 40
matches: 50
matches: 41
matches: 37
matches: 44
matches: 43
matches: 41
matches: 41
matches: 51
matches: 30
matches: 38
matches: 45
matches: 34
matches: 34
matches: 47
matches: 42
matches: 41
matches: 49
matches: 44
matches: 36
matches: 46
matches: 42
matches: 34
matches: 37
matches: 35
matches: 37
matches: 42
matches: 38
matches: 52
matches: 38
matches: 37
matches: 40
matches: 39
matches: 47
matches: 42
matches: 31
matches: 34
matches: 39
matches: 35
matches: 46
matches: 36
matches: 28
matches: 41
matches: 31
matches: 37
matches: 38
matches: 46
matches: 42
matches: 41
matches: 42
matches: 44
matches: 39
matches: 30
matches: 35
matches: 36
matches: 35
matches: 51
matches: 40
matches: 33
matches: 34
matches: 45
matches: 31
matches: 50
matches: 40
matches: 35
matches: 39
matches: 37
matches: 37
matches: 41
matches: 36
matches: 36
matches: 36
matches: 50
matches: 41
matches: 48
matches: 45
matches: 36
matches: 38
matches: 52
matches: 40
matches: 43
matches: 43
matches: 49
matches: 34
matches: 32
matches: 35
matches: 40
matches: 30
matches: 44
matches: 29
matches: 48
matches: 41
matches: 37
matches: 44
matches: 43
matches: 37
matches: 44
matches: 42
matches: 39
matches: 41
matches: 42
matches: 39
matches: 43
matches: 36
matches: 45
matches: 44
matches: 34
matches: 50
matches: 47
matches: 46
matches: 42
matches: 43
matches: 40
matches: 35
matches: 38
matches: 43
matches: 45
matches: 34
matches: 47
matches: 45
matches: 55
matches: 35
matches: 39
matches: 50
matches: 42
matches: 38
matches: 40
matches: 46
matches: 52
matches: 42
matches: 43
matches: 38
matches: 42
matches: 45
matches: 43
matches: 27
matches: 39
matches: 37
matches: 35
matches: 35
matches: 44
matches: 36
matches: 34
matches: 41
matches: 37
matches: 39
matches: 33
matches: 30
matches: 38
matches: 42
matches: 34
matches: 41
matches: 38
matches: 49
matches: 40
matches: 43
matches: 49
matches: 42
matches: 44
matches: 40
matches: 51
matches: 38
matches: 35
matches: 48
matches: 36
matches: 47
matches: 39
matches: 33
matches: 34
matches: 30
matches: 36
matches: 36
matches: 44
matches: 39
matches: 34
matches: 34
matches: 49
matches: 42
matches: 40
matches: 39
matches: 36
matches: 36
matches: 45
matches: 39
matches: 35
matches: 30
matches: 45
matches: 37
matches: 42
matches: 39
matches: 36
matches: 41
matches: 33
matches: 41
matches: 34
matches: 41
matches: 43
matches: 41
matches: 32
matches: 49
matches: 32
matches: 45
matches: 40
matches: 37
matches: 43
matches: 50
matches: 42
matches: 42
matches: 35
matches: 40
matches: 36
matches: 30
matches: 40
matches: 44
matches: 54
matches: 37
matches: 46
matches: 41
matches: 35
matches: 55
matches: 38
matches: 32
matches: 34
matches: 49
matches: 40
matches: 34
matches: 40
matches: 33
matches: 49
matches: 33
matches: 40
matches: 39
matches: 39
matches: 35
matches: 41
matches: 52
matches: 43
matches: 44
matches: 43
matches: 43
matches: 38
matches: 37
matches: 43
matches: 45
matches: 41
matches: 44
matches: 38
matches: 38
matches: 41
matches: 29
matches: 30
matches: 40
matches: 41
matches: 37
matches: 46
matches: 44
matches: 36
matches: 37
matches: 38
matches: 34
matches: 49
matches: 45
matches: 42
matches: 37
matches: 37
matches: 44
matches: 43
matches: 44
matches: 44
matches: 40
matches: 38
matches: 39
matches: 41
matches: 41
matches: 39
matches: 41
matches: 30
matches: 33
matches: 38
matches: 41
matches: 42
matches: 43
matches: 43
matches: 44
matches: 38
matches: 42
matches: 38
matches: 41
matches: 47
matches: 41
matches: 43
matches: 46
matches: 34
matches: 33
matches: 41
matches: 44
matches: 40
matches: 52
matches: 37
matches: 38
matches: 32
matches: 47
matches: 36
matches: 39
matches: 50
matches: 37
matches: 40
matches: 33
matches: 50
matches: 52
matches: 43
matches: 41
matches: 44
matches: 49
matches: 38
matches: 38
matches: 43
matches: 42
matches: 45
matches: 34
matches: 39
matches: 48
matches: 42
matches: 48
matches: 40
matches: 49
matches: 29
matches: 37
matches: 36
matches: 32
matches: 40
matches: 39
matches: 43
matches: 35
matches: 45
matches: 33
matches: 43
matches: 42
matches: 30
matches: 39
matches: 34
matches: 42
matches: 41
matches: 42
matches: 37
matches: 41
matches: 34
matches: 31
matches: 38
matches: 37
matches: 44
matches: 38
matches: 30
matches: 45
matches: 53
matches: 49
matches: 39
matches: 37
matches: 47
matches: 48
matches: 50
matches: 44
matches: 32
matches: 48
matches: 56
matches: 46
matches: 40
matches: 45
matches: 40
matches: 47
matches: 35
matches: 38
matches: 40
matches: 46
matches: 44
matches: 32
matches: 47
matches: 30
matches: 36
matches: 33
matches: 40
matches: 35
matches: 31
matches: 39
matches: 34
matches: 43
matches: 34
matches: 26
matches: 35
matches: 36
matches: 41
matches: 31
matches: 42
matches: 44
matches: 36
matches: 53
matches: 35
matches: 38
matches: 42
matches: 36
matches: 47
matches: 38
matches: 44
matches: 37
matches: 47
matches: 36
matches: 32
matches: 31
matches: 40
matches: 42
matches: 44
matches: 35
matches: 30
matches: 40
matches: 46
matches: 44
matches: 43
matches: 41
matches: 37
matches: 33
matches: 42
matches: 45
matches: 47
matches: 37
matches: 41
matches: 38
matches: 41
matches: 39
matches: 29
matches: 44
matches: 48
matches: 43
matches: 39
matches: 41
matches: 29
matches: 40
matches: 51
matches: 42
matches: 43
matches: 41
matches: 40
matches: 41
matches: 44
matches: 42
matches: 40
matches: 48
matches: 43
matches: 36
matches: 35
matches: 47
matches: 35
matches: 45
matches: 45
matches: 40
matches: 43
matches: 27
matches: 41
matches: 38
matches: 34
matches: 34
matches: 46
matches: 29
matches: 36
matches: 40
matches: 38
matches: 37
matches: 50
matches: 37
matches: 40
matches: 47
matches: 39
matches: 43
matches: 46
matches: 38
matches: 41
matches: 41
matches: 44
matches: 38
matches: 38
matches: 39
matches: 42
matches: 42
matches: 32
matches: 36
matches: 38
matches: 36
matches: 36
matches: 49
matches: 46
matches: 40
matches: 32
matches: 37
matches: 43
matches: 32
matches: 42
matches: 34
matches: 33
matches: 44
matches: 44
matches: 36
matches: 30
matches: 45
matches: 31
matches: 46
matches: 32
matches: 35
matches: 30
matches: 45
matches: 36
matches: 45
matches: 39
matches: 36
matches: 56
matches: 43
matches: 39
matches: 39
matches: 38
matches: 39
matches: 40
matches: 47
matches: 40
matches: 41
matches: 36
matches: 40
matches: 41
matches: 55
matches: 49
matches: 45
matches: 38
matches: 31
matches: 39
matches: 39
matches: 31
matches: 37
matches: 29
matches: 32
matches: 37
matches: 33
matches: 36
matches: 48
matches: 38
matches: 47
matches: 38
matches: 43
matches: 46
matches: 54
matches: 34
matches: 27
matches: 47
matches: 40
matches: 39
matches: 34
matches: 39
matches: 46
matches: 39
matches: 35
matches: 48
matches: 47
matches: 42
matches: 30
matches: 41
matches: 34
matches: 47
matches: 47
matches: 34
matches: 48
matches: 37
matches: 44
matches: 42
matches: 44
matches: 32
matches: 29
matches: 44
matches: 42
matches: 36
matches: 48
matches: 35
matches: 38
matches: 46
matches: 39
matches: 43
matches: 41
matches: 54
matches: 47
matches: 32
matches: 39
matches: 36
matches: 38
matches: 43
matches: 48
matches: 33
matches: 39
matches: 44
matches: 35
matches: 40
matches: 50
matches: 48
matches: 31
matches: 37
matches: 51
matches: 47
matches: 35
matches: 47
matches: 43
matches: 44
matches: 40
matches: 33
matches: 46
matches: 41
matches: 43
matches: 38
matches: 38
matches: 40
matches: 41
matches: 35
matches: 44
matches: 47
matches: 36
matches: 45
matches: 41
matches: 42
matches: 47
matches: 43
matches: 37
matches: 39
matches: 35
matches: 36
matches: 44
matches: 39
matches: 40
matches: 41
matches: 46
matches: 38
matches: 34
matches: 33
matches: 42
matches: 40
matches: 40
matches: 39
matches: 53
matches: 42
matches: 45
matches: 35
matches: 41
matches: 39
matches: 39
matches: 44
matches: 36
matches: 47
matches: 40
matches: 38
matches: 43
matches: 43
matches: 42
matches: 38
matches: 47
matches: 32
matches: 33
matches: 38
matches: 50
matches: 53
matches: 34
matches: 48
matches: 44
matches: 44
matches: 35
matches: 38
matches: 54
matches: 33
matches: 37
matches: 41
matches: 31
matches: 33
matches: 34
matches: 35
matches: 47
matches: 41
matches: 43
matches: 36
matches: 41
matches: 42
matches: 42
matches: 49
matches: 35
matches: 43
matches: 36
matches: 48
matches: 39
matches: 41
matches: 42
matches: 38
matches: 46
matches: 47
matches: 37
matches: 39
matches: 44
matches: 40
matches: 44
matches: 45
matches: 41
matches: 44
matches: 46
matches: 37
matches: 41
matches: 48
matches: 37
matches: 39
matches: 38
matches: 28
matches: 45
matches: 48
matches: 41
matches: 33
matches: 37
matches: 41
matches: 44
matches: 32
matches: 47
matches: 32
matches: 40
matches: 38
matches: 44
matches: 39
matches: 46
matches: 39
matches: 36
matches: 39
matches: 47
matches: 32
matches: 42
matches: 35
matches: 40
matches: 47
matches: 48
matches: 30
matches: 42
matches: 40
matches: 35
matches: 41
matches: 41
matches: 38
matches: 37
matches: 48
matches: 46
matches: 26
matches: 33
matches: 37
matches: 49
matches: 47
matches: 36
matches: 31
matches: 20
matches: 39
matches: 34
matches: 40
matches: 45
matches: 33
matches: 31
matches: 33
matches: 38
matches: 42
matches: 36
matches: 39
matches: 46
matches: 38
matches: 48
matches: 44
matches: 36
matches: 41
matches: 41
matches: 44
matches: 49
matches: 48
matches: 40
matches: 41
matches: 45
matches: 33
matches: 41
matches: 36
matches: 42
matches: 44
matches: 40
matches: 40
matches: 46
matches: 42
matches: 33
matches: 37
matches: 48
matches: 39
matches: 40
matches: 39
matches: 43
matches: 33
matches: 36
matches: 44
matches: 36
matches: 42
matches: 49
matches: 43
matches: 38
matches: 42
matches: 49
matches: 38
matches: 35
matches: 43
matches: 48
matches: 33
matches: 39
matches: 30
matches: 32
matches: 43
matches: 35
matches: 39
matches: 44
matches: 35
matches: 37
matches: 34
matches: 41
matches: 29
matches: 34
matches: 36
matches: 33
matches: 40
matches: 36
matches: 46
matches: 46
matches: 38
matches: 42
matches: 42
matches: 46
matches: 44
matches: 42
matches: 42
matches: 39
matches: 37
matches: 37
matches: 29
matches: 35
matches: 45
matches: 37
matches: 38
matches: 38
matches: 42
matches: 29
matches: 43
matches: 51
matches: 46
matches: 39
matches: 41
matches: 29
matches: 41
matches: 51
matches: 36
matches: 45
matches: 48
matches: 54
matches: 48
matches: 45
matches: 41
matches: 40
matches: 43
c:\users\aparn\anaconda\envs\my_first_env\lib\site-packages\ipykernel\__main__.py:30: SettingWithCopyWarning:
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead
See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy
In [5]:
for col in c.columns:
print(col)
ltable_id
rtable_id
ltable_name
ltable_genus
ltable_family
ltable_ecology
ltable_countries
ltable_threat_paragraph
ltable_conservation_paragraph
ltable_pop_trend
ltable_status
ltable_country_count
ltable_scientific_name
rtable_scientific_name
rtable_name
rtable_kingdom
rtable_phylum
rtable_class
rtable_order
rtable_family
rtable_genus
rtable_size
rtable_threats
rtable_conservation
rtable_threat_keywords
rtable_conservation_keywords
rtable_status
rtable_countries
rtable_country_count
rtable_tCount
_id.1
ltable_id.1
rtable_id.1
name_name_jac_qgm_3_qgm_3
name_name_jac_dlm_dc0_dlm_dc0
countries_countries_jac_qgm_3_qgm_3
countries_countries_cos_dlm_dc0_dlm_dc0
countries_countries_jac_dlm_dc0_dlm_dc0
countries_countries_mel
countries_countries_lev_dist
countries_countries_lev_sim
countries_countries_nmw
countries_countries_sw
country_count_country_count_exm
country_count_country_count_anm
country_count_country_count_lev_dist
country_count_country_count_lev_sim
status_match
country_overlap
country_count_sim
In [6]:
# gold.to_csv('./gold.csv')
gold.to_csv('./gold.csv')
# unlabeled.to_csv('./unlabeled.csv')
unlabeled.to_csv('./unlabeled.csv')
In [7]:
# remove unnecessary features
# write out gold and unlabeled csvs with only selected features for matcher
#cols_to_keep = ['country_count_sim',
# 'status_match',
# 'country_overlap',
# 'name_name_jac_dlm_dc0_dlm_dc0',
# 'name_name_jac_qgm_3_qgm_3',
# 'countries_countries_jac_dlm_dc0_dlm_dc0',
# 'countries_countries_jac_qgm_3_qgm_3'
# ]
#cols_to_keep version 2
cols_to_keep = ['name_name_jac_qgm_3_qgm_3',
'name_name_jac_dlm_dc0_dlm_dc0',
'countries_countries_jac_qgm_3_qgm_3',
'countries_countries_cos_dlm_dc0_dlm_dc0',
'countries_countries_jac_dlm_dc0_dlm_dc0',
'countries_countries_mel',
'countries_countries_lev_dist',
'countries_countries_lev_sim',
'countries_countries_nmw',
'countries_countries_sw',
'country_count_country_count_exm',
'country_count_country_count_anm',
'country_count_country_count_lev_dist',
'country_count_country_count_lev_sim',
'status_match',
'country_overlap',
'country_count_sim']
# FOR NOW, just use these while we wait for other features
# cols_to_keep_gold = ['country_count_sim',
# 'name_name_jac_dlm_dc0_dlm_dc0',
# 'name_name_jac_qgm_3_qgm_3',
# 'label']
gold_reduced = gold[cols_to_keep + ['label']]
# cols_to_keep_unlabeled = ['country_count_sim',
# 'name_name_jac_dlm_dc0_dlm_dc0',
# 'name_name_jac_qgm_3_qgm_3']
unlabeled_reduced = unlabeled[cols_to_keep]
In [8]:
# gold_reduced.to_csv('./gold_reduced.csv')
gold_reduced.to_csv('./gold_reduced.csv')
# unlabeled_reduced.to_csv('./unlabeled_reduced.csv')
unlabeled_reduced.to_csv('./unlabeled_reduced.csv')
In [9]:
#using features for matcher
train,test = train_test_split(gold, train_size=0.7, test_size=0.3)
len(train)
train_reduced = train[cols_to_keep + ['label']]
test_reduced = test[cols_to_keep]
In [10]:
train.to_csv('./train.csv')
test.to_csv('./test.csv')
train_reduced.to_csv('./train_reduced.csv')
test_reduced.to_csv('./test_reduced.csv')
In [12]:
train_reduced
Out[12]:
name_name_jac_qgm_3_qgm_3
name_name_jac_dlm_dc0_dlm_dc0
countries_countries_jac_qgm_3_qgm_3
countries_countries_cos_dlm_dc0_dlm_dc0
countries_countries_jac_dlm_dc0_dlm_dc0
countries_countries_mel
countries_countries_lev_dist
countries_countries_lev_sim
countries_countries_nmw
countries_countries_sw
country_count_country_count_exm
country_count_country_count_anm
country_count_country_count_lev_dist
country_count_country_count_lev_sim
status_match
country_overlap
country_count_sim
label
_id
7281363
NaN
NaN
0.076923
0.0
0.0
0.483487
26
0.161290
-19.0
5.0
0
0.500000
1.0
0.0
1
1.0
1
0
7278036
NaN
NaN
0.000000
0.0
0.0
0.431900
28
0.096774
-22.0
2.0
0
0.500000
1.0
0.0
0
0.0
1
0
2041033
NaN
NaN
0.000000
0.0
0.0
0.441667
15
0.062500
-5.0
2.0
1
1.000000
0.0
1.0
0
0.0
1
0
4034499
0.285714
0.333333
0.000000
0.0
0.0
0.433211
31
0.114286
-18.0
2.0
0
0.500000
1.0
0.0
1
0.0
1
0
2029105
NaN
NaN
0.000000
0.0
0.0
0.383572
39
0.093023
-26.0
1.0
1
1.000000
0.0
1.0
1
0.0
1
0
1369586
NaN
NaN
0.000000
0.0
0.0
0.513095
33
0.153846
0.0
4.0
0
0.333333
1.0
0.0
1
0.0
0
0
4992441
NaN
NaN
0.000000
0.0
0.0
0.443254
21
0.125000
-6.0
2.0
0
0.500000
1.0
0.0
0
0.0
1
0
5992303
NaN
NaN
0.000000
0.0
0.0
0.455556
11
0.083333
-1.0
1.0
1
1.000000
0.0
1.0
0
0.0
1
0
4414743
0.068966
0.250000
0.018519
0.0
0.0
0.549383
51
0.055556
-42.0
3.0
0
0.500000
1.0
0.0
0
0.0
1
0
2103792
0.206897
0.250000
0.000000
0.0
0.0
0.379426
35
0.102564
-21.0
2.0
0
0.500000
1.0
0.0
0
0.0
1
0
1353765
NaN
NaN
0.000000
0.0
0.0
0.397547
30
0.090909
-23.0
1.0
0
0.333333
1.0
0.0
1
0.0
0
0
5251547
NaN
NaN
0.000000
0.0
0.0
0.000000
12
0.000000
-4.0
0.0
1
1.000000
0.0
1.0
1
0.0
1
0
5048744
NaN
NaN
0.000000
0.0
0.0
0.000000
13
0.000000
-11.0
0.0
0
0.000000
1.0
0.0
0
0.0
1
0
5843893
NaN
NaN
0.000000
0.0
0.0
0.410109
23
0.080000
0.0
2.0
0
0.500000
1.0
0.0
0
0.0
1
0
8681483
NaN
NaN
0.000000
0.0
0.0
0.417989
9
0.000000
-2.0
1.0
1
1.000000
0.0
1.0
1
0.0
1
1
2350777
0.310345
0.285714
0.000000
0.0
0.0
0.440571
104
0.045872
-94.0
2.0
0
0.111111
1.0
0.0
1
0.0
0
1
1369592
NaN
NaN
0.000000
0.0
0.0
0.576263
30
0.090909
0.0
4.0
0
0.333333
1.0
0.0
1
0.0
0
0
2111714
NaN
NaN
0.019231
0.0
0.0
0.401641
27
0.181818
-7.0
3.0
0
0.333333
1.0
0.0
0
0.0
0
0
5661990
NaN
NaN
0.000000
0.0
0.0
0.426244
14
0.176471
-1.0
2.0
1
1.000000
0.0
1.0
1
0.0
1
0
7692953
0.352941
0.500000
0.000000
0.0
0.0
0.390756
30
0.117647
-23.0
2.0
0
0.333333
1.0
0.0
1
0.0
0
0
4833596
0.257143
0.500000
0.045455
0.0
0.0
0.500174
19
0.240000
-3.0
4.0
0
0.500000
1.0
0.0
0
0.0
1
0
9148925
NaN
NaN
0.000000
0.0
0.0
0.381481
28
0.066667
-10.0
2.0
1
1.000000
0.0
1.0
1
0.0
1
0
6261720
NaN
NaN
0.000000
0.0
0.0
0.000000
17
0.000000
-15.0
0.0
0
0.000000
1.0
0.0
0
0.0
0
0
8160094
0.196078
0.285714
0.000000
0.0
0.0
0.463297
21
0.160000
-8.0
3.0
1
1.000000
0.0
1.0
1
0.0
1
0
2103735
0.071429
0.250000
0.000000
0.0
0.0
0.406136
12
0.142857
1.0
2.0
1
1.000000
0.0
1.0
0
0.0
1
0
5517359
0.800000
1.000000
0.000000
0.0
0.0
0.300239
38
0.116279
-25.0
2.0
0
0.333333
1.0
0.0
1
0.0
0
1
7817848
0.147059
0.333333
0.000000
0.0
0.0
0.487037
13
0.133333
-4.0
1.0
1
1.000000
0.0
1.0
1
0.0
1
0
4414610
NaN
NaN
0.000000
0.0
0.0
0.368687
10
0.090909
-1.0
1.0
1
1.000000
0.0
1.0
0
0.0
1
0
4416982
NaN
NaN
0.000000
0.0
0.0
0.301471
97
0.049020
-89.0
2.0
0
0.111111
1.0
0.0
1
0.0
0
0
2103719
0.000000
0.000000
0.000000
0.0
0.0
0.406136
12
0.142857
1.0
2.0
1
1.000000
0.0
1.0
0
0.0
1
0
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
7274994
NaN
NaN
0.441860
0.0
0.0
0.887694
7
0.774194
20.0
22.0
1
1.000000
0.0
1.0
0
1.0
1
0
7692954
NaN
NaN
0.000000
0.0
0.0
0.464052
32
0.058824
-26.0
1.0
0
0.333333
1.0
0.0
1
0.0
0
0
8429089
0.527778
0.500000
0.000000
0.0
0.0
0.000000
12
0.000000
-10.0
0.0
0
0.000000
1.0
0.0
1
0.0
1
1
4889658
NaN
NaN
0.027778
0.0
0.0
0.293590
19
0.050000
-6.0
3.0
1
1.000000
0.0
1.0
0
0.0
1
0
7278093
NaN
NaN
0.410256
0.0
0.0
0.854301
12
0.612903
8.0
18.0
0
0.500000
1.0
0.0
0
1.0
1
0
6181282
0.195652
0.250000
0.000000
0.0
0.0
0.446635
39
0.113636
-22.0
3.0
0
0.333333
1.0
0.0
1
0.0
0
0
4889643
0.000000
0.000000
0.134615
0.0
0.0
0.619658
47
0.129630
-36.0
9.0
0
0.500000
1.0
0.0
0
1.0
1
0
7689612
NaN
NaN
0.000000
0.0
0.0
0.376634
31
0.088235
-23.0
2.0
0
0.333333
1.0
0.0
0
0.0
0
0
2107013
NaN
NaN
0.000000
0.0
0.0
0.492424
29
0.121212
-18.0
2.0
0
0.333333
1.0
0.0
1
0.0
0
0
463646
0.800000
1.000000
0.000000
0.0
0.0
0.488154
58
0.147059
-34.0
5.0
0
0.600000
1.0
0.0
1
0.0
0
1
75598
0.791667
1.000000
0.351351
0.0
0.0
0.531032
21
0.160000
-2.0
10.0
1
1.000000
0.0
1.0
1
1.0
1
1
4979396
0.171429
0.250000
0.000000
0.0
0.0
0.378788
20
0.090909
-11.0
1.0
1
1.000000
0.0
1.0
1
0.0
1
0
8876418
NaN
NaN
0.000000
0.0
0.0
0.000000
27
0.000000
-25.0
0.0
0
0.000000
1.0
0.0
0
0.0
1
0
4885560
0.208333
0.333333
0.072917
0.0
0.0
0.427139
92
0.089109
-79.0
9.0
0
0.333333
1.0
0.0
1
1.0
0
0
1356294
NaN
NaN
0.000000
0.0
0.0
0.407828
30
0.090909
-14.0
3.0
0
0.333333
1.0
0.0
1
0.0
0
0
3407725
0.133333
0.000000
0.000000
0.0
0.0
0.455556
11
0.083333
-1.0
1.0
1
1.000000
0.0
1.0
1
0.0
1
0
7273809
0.090909
0.111111
0.000000
0.0
0.0
0.349145
19
0.050000
-6.0
2.0
0
0.500000
1.0
0.0
1
0.0
1
0
2111603
NaN
NaN
0.000000
0.0
0.0
0.471026
29
0.121212
-10.0
3.0
0
0.666667
1.0
0.0
0
0.0
1
0
5251546
NaN
NaN
0.000000
0.0
0.0
0.403595
14
0.176471
-2.0
1.0
0
0.500000
1.0
0.0
1
0.0
1
1
2099476
NaN
NaN
0.000000
0.0
0.0
0.424162
47
0.129630
-33.0
2.0
0
0.500000
1.0
0.0
1
0.0
1
0
2520162
0.600000
0.500000
0.025000
0.0
0.0
0.509187
19
0.173913
-4.0
4.0
0
0.500000
1.0
0.0
1
0.0
1
1
2103785
NaN
NaN
0.000000
0.0
0.0
0.379426
35
0.102564
-21.0
2.0
0
0.500000
1.0
0.0
0
0.0
1
0
1817922
0.200000
0.333333
0.000000
0.0
0.0
0.382479
11
0.153846
0.0
3.0
1
1.000000
0.0
1.0
0
0.0
1
0
8267058
0.263158
0.400000
0.000000
0.0
0.0
0.477273
9
0.181818
-1.0
1.0
1
1.000000
0.0
1.0
0
0.0
1
0
4944347
0.181818
0.333333
0.000000
0.0
0.0
0.455556
11
0.083333
-1.0
1.0
1
1.000000
0.0
1.0
0
0.0
1
0
2111632
0.761905
1.000000
0.014286
0.0
0.0
0.463453
31
0.205128
0.0
6.0
0
0.666667
1.0
0.0
0
0.0
1
0
7692975
0.406250
0.500000
0.000000
0.0
0.0
0.390756
30
0.117647
-23.0
2.0
0
0.333333
1.0
0.0
1
0.0
0
0
4414761
NaN
NaN
0.019231
0.0
0.0
0.556397
41
0.068182
-32.0
3.0
0
0.500000
1.0
0.0
0
0.0
1
0
6064557
NaN
NaN
0.000000
0.0
0.0
0.500000
12
0.200000
-2.0
2.0
1
1.000000
0.0
1.0
0
0.0
1
1
8884319
NaN
NaN
0.000000
0.0
0.0
0.451852
14
0.066667
-5.0
1.0
1
1.000000
0.0
1.0
0
0.0
1
0
279 rows × 18 columns
In [ ]:
Content source: andrewedstrom/cs638project
Similar notebooks: