In [1]:

    

import pandas as pd
import matplotlib.pyplot as plt


    

In [2]:

    

train_set_path = r'C:\Users\dmpas\thesis\data\network\bitcoin\estimates\train_set.csv'
test_set_path = r'C:\Users\dmpas\thesis\data\network\bitcoin\estimates\test_set.csv'


    

In [3]:

    

def ma(price):
    if price < -100:
        return 5
    if price < -50:
        return 4
    if price < 0:
        return 3
    if price < 50:
        return 2
    return 1


    

In [4]:

    

train_df = pd.read_csv(train_set_path, index_col=0)
train_df.columns = ['Author', 'bc', 'cc', 'chibs', 'dc', 'ec', 'hm', 'is', 'lr', 'price']
train_df['price_category'] = train_df.price.map(ma)

print(len(train_df))


    

    




356557

In [5]:

    

train_df.head()


    

    
Out[5]:







  

    

      

      
Author
      
bc
      
cc
      
chibs
      
dc
      
ec
      
hm
      
is
      
lr
      
price
      
price_category
    
  
  

    

      
0
      
177412
      
0.634528
      
1.239146
      
-18.257680
      
1.438233
      
0.372907
      
-1325.966306
      
-22076.378149
      
-27003.578634
      
6.9
      
2
    
    

      
1
      
53511
      
0.262976
      
1.225784
      
-101.154828
      
1.062802
      
0.305459
      
-498.691400
      
-7557.808217
      
-7461.446531
      
6.9
      
2
    
    

      
2
      
156757
      
0.091405
      
1.200112
      
-3.394149
      
0.757074
      
0.209725
      
-237.573484
      
-3396.268925
      
-4125.309848
      
6.9
      
2
    
    

      
3
      
100398
      
0.098763
      
1.190790
      
-26.896536
      
0.670117
      
0.167796
      
-252.269969
      
-3612.483019
      
-3672.512010
      
6.9
      
2
    
    

      
4
      
166687
      
0.065896
      
1.192045
      
-8.781838
      
0.503796
      
0.157702
      
-117.802576
      
-1858.142623
      
-1775.790037
      
6.9
      
2
    
  

In [6]:

    

test_df = pd.read_csv(test_set_path, index_col=0)
test_df.columns = ['Author', 'bc', 'cc', 'chibs', 'dc', 'ec', 'hm', 'is', 'lr', 'price']
test_df['price_category'] = train_df.price.map(ma)

print(len(test_df))


    

    




88521

In [7]:

    

test_df.head()


    

    
Out[7]:







  

    

      

      
Author
      
bc
      
cc
      
chibs
      
dc
      
ec
      
hm
      
is
      
lr
      
price
      
price_category
    
  
  

    

      
0
      
99508
      
0.022297
      
1.169596
      
-30.870187
      
0.502415
      
0.150627
      
-222.525649
      
-2995.403068
      
-2606.443962
      
6.9
      
2
    
    

      
1
      
177268
      
0.016294
      
1.159342
      
-9.926871
      
0.429262
      
0.128385
      
-73.944996
      
-1053.014603
      
-954.976211
      
6.9
      
2
    
    

      
2
      
50584
      
0.017190
      
1.161668
      
16.082362
      
0.324362
      
0.094380
      
-84.019078
      
-1102.406890
      
-1055.546625
      
6.9
      
2
    
    

      
3
      
153645
      
0.008935
      
1.117261
      
-12.029111
      
0.314010
      
0.091849
      
-86.938191
      
-981.997684
      
-975.776543
      
6.9
      
2
    
    

      
4
      
250893
      
0.004391
      
1.115628
      
-20.254224
      
0.287785
      
0.085615
      
-34.020792
      
-442.435106
      
-418.764624
      
6.9
      
2
    
  

In [8]:

    

from sklearn import preprocessing


    

In [9]:

    

train_set_X = preprocessing.scale(train_df.loc[:, ['bc', 'cc', 'chibs', 'dc', 'ec', 'hm', 'is', 'lr']])
train_set_Y = train_df.loc[:, ['price_category']].values.ravel()


    

In [10]:

    

test_set_X = preprocessing.scale(test_df.loc[:, ['bc', 'cc', 'chibs', 'dc', 'ec', 'hm', 'is', 'lr']])
test_set_Y = test_df.loc[:, ['price_category']].values.ravel()


    

In [11]:

    

from sklearn.ensemble import BaggingClassifier, RandomForestClassifier
from sklearn.multiclass import OneVsRestClassifier
from sklearn.metrics import classification_report, confusion_matrix
from sklearn.svm import SVC


    

In [12]:

    

n_estimators = 8


    

In [13]:

    

svc1 = OneVsRestClassifier(BaggingClassifier(SVC(kernel='linear'), max_samples=1.0/50, n_estimators=n_estimators))
svc1.fit(train_set_X, train_set_Y)

y_pred_1 = svc1.predict(test_set_X)
print(confusion_matrix(test_set_Y, y_pred_1))


    

    




[[ 7885 18471   371  4269   144]
 [ 4817 12209   291  2016   118]
 [ 2485  7860    77  1930    34]
 [    0     0     0     0     0]
 [ 6564 14908   305  3663   104]]

In [ ]:

    




    

In [14]:

    

svc2 = OneVsRestClassifier(BaggingClassifier(SVC(kernel='poly', degree=4), max_samples=1.0/50, n_estimators=n_estimators))
svc2.fit(train_set_X, train_set_Y)
y_pred_2 = svc2.predict(test_set_X)

print(confusion_matrix(test_set_Y, y_pred_2))


    

    




[[18453  3784  2822  2297  3784]
 [11236  2629  1793  1355  2438]
 [ 7944  1293   879  1028  1242]
 [    0     0     0     0     0]
 [16216  2626  1907  1833  2962]]

In [ ]:

    




    

In [15]:

    

svc4 = OneVsRestClassifier(BaggingClassifier(SVC(kernel='rbf'), max_samples=1.0/50, n_estimators=n_estimators))
svc4.fit(train_set_X, train_set_Y)
y_pred_4 = svc4.predict(test_set_X)

print(confusion_matrix(test_set_Y, y_pred_4))


    

    




[[14610  5656  2322  2365  6187]
 [ 9854  3647  1438  1193  3319]
 [ 5580  1684   965  1338  2819]
 [    0     0     0     0     0]
 [12206  3547  1881  2302  5608]]

In [ ]:

    




    

In [16]:

    

svc5 = OneVsRestClassifier(BaggingClassifier(SVC(kernel='sigmoid'), max_samples=1.0/50, n_estimators=n_estimators))
svc5.fit(train_set_X, train_set_Y)
y_pred_5 = svc5.predict(test_set_X)

print(confusion_matrix(test_set_Y, y_pred_5))


    

    




[[26216  1988     0  2827   109]
 [16901  1239     0  1263    48]
 [10551   766     0  1018    51]
 [    0     0     0     0     0]
 [21397  1648     0  2406    93]]

In [ ]:

    




    

In [17]:

    

rfc = RandomForestClassifier(min_samples_leaf=20)
rfc.fit(train_set_X, train_set_Y)

y_pred_6 = rfc.predict(test_set_X)
print(confusion_matrix(test_set_Y, y_pred_6))


    

    




[[13250  7652  5042  2703  2493]
 [ 9090  5503  1906  2203   749]
 [ 5503  1850  1091  2625  1317]
 [    0     0     0     0     0]
 [15069  4226  1480  2500  2269]]

In [ ]:

    




    

In [ ]: