In [1]:

    

import pandas as pd
import numpy as np
from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import train_test_split
from sklearn import metrics


    

In [2]:

    

base_path = "/Users/qiangzhang/Google Drive/workspace (1)/Agent-analysis/data/"
parcel_path = base_path + "Regression.csv"


    

In [3]:

    

data = pd.read_csv(parcel_path)
data['sold'] = (data.sold > 0).astype(int)


    

In [ ]:

    




    

In [4]:

    

data.groupby('sold').mean()


    

    
Out[4]:






  

    

      

      
Unnamed: 0
      
PIN
      
age
      
price
      
salary
      
year
    
    

      
sold
      

      

      

      

      

      

    
  
  

    

      
0
      
54347.107916
      
5.352586e+13
      
51.571959
      
106475.372446
      
59388.106064
      
2013.995865
    
    

      
1
      
56861.829400
      
5.372249e+13
      
52.434077
      
289176.123043
      
58641.998376
      
2014.083568
    
  

In [5]:

    

data.groupby('year').mean()


    

    
Out[5]:






  

    

      

      
Unnamed: 0
      
PIN
      
age
      
price
      
salary
      
sold
    
    

      
year
      

      

      

      

      

      

    
  
  

    

      
2013
      
18107.5
      
5.353428e+13
      
50.627043
      
114177.364995
      
59256.484123
      
0.075574
    
    

      
2014
      
54380.5
      
5.354174e+13
      
51.682191
      
113424.259400
      
59012.693752
      
0.082356
    
    

      
2015
      
90833.0
      
5.355143e+13
      
52.615858
      
138092.627642
      
59702.884101
      
0.095940
    
  

In [6]:

    

data_sold = data.loc[data['sold'] == 1]
data_unsold = data.loc[data['sold'] == 0]


    

In [7]:

    

#Train size = 2000, test size = 2000
#features = ['age','salary','sold price']
XS_train, XS_test, yS_train, yS_test = train_test_split(data_sold[['age','salary','price']], data_sold['sold'], train_size=2000, test_size=2000,random_state=42)
XU_train, XU_test, yU_train, yU_test = train_test_split(data_unsold[['age','salary','price']], data_unsold['sold'], train_size=2000, test_size=2000,random_state=42)

X_train = XS_train.append(XU_train).values
X_test = XS_test.append(XU_test).values
y_train = yS_train.append(yU_train).values
y_test = yS_test.append(yU_test).values


    

In [8]:

    

model = LogisticRegression()
model = model.fit(X_train, y_train)

print model.score(X_train, y_train)


    

    




0.67875

In [9]:

    

print y_train.mean()


    

    




0.5

In [10]:

    

print "age, salary, price coefficients"
print model.coef_


    

    




age, salary, price coefficients
[[ -1.77362350e-08  -1.97805758e-05   7.88090537e-06]]

In [11]:

    

predicted = model.predict(X_test)
probs = model.predict_proba(X_test)


    

In [12]:

    

# generate class probabilities
print probs


    

    




[[ 0.59993621  0.40006379]
 [ 0.25368794  0.74631206]
 [ 0.24158786  0.75841214]
 ..., 
 [ 0.34754046  0.65245954]
 [ 0.62314857  0.37685143]
 [ 0.36337563  0.63662437]]

In [13]:

    

# generate evaluation metrics
print metrics.accuracy_score(y_test, predicted)
print metrics.roc_auc_score(y_test, probs[:, 1])


    

    




0.6825
0.73320175

In [14]:

    

print metrics.classification_report(y_test, predicted)


    

    




             precision    recall  f1-score   support

          0       0.66      0.74      0.70      2000
          1       0.71      0.62      0.66      2000

avg / total       0.69      0.68      0.68      4000

In [15]:

    

print metrics.confusion_matrix(y_test, predicted)


    

    




[[1485  515]
 [ 755 1245]]

In [ ]:

    




    

In [ ]: