Gold Proximity in India

In [1]:

    

import numpy as np
import pandas as pd
%pylab inline
pylab.style.use('ggplot')


    

    




Populating the interactive namespace from numpy and matplotlib

In [56]:

    

url  = 'http://www.stat.ufl.edu/~winner/data/gold_target1.dat'
data_df = pd.read_csv(url, sep='[\s]+', engine='python', header=None)


    

In [57]:

    

data_df.info()


    

    




<class 'pandas.core.frame.DataFrame'>
RangeIndex: 64 entries, 0 to 63
Data columns (total 4 columns):
0    64 non-null float64
1    64 non-null float64
2    64 non-null int64
3    64 non-null int64
dtypes: float64(2), int64(2)
memory usage: 2.1 KB

In [58]:

    

data_df.columns = ['as_level', 'sb_level', 'l_proximity', 'gold_proximity']


    

In [59]:

    

data_df.head()


    

    
Out[59]:






  

    

      

      
as_level
      
sb_level
      
l_proximity
      
gold_proximity
    
  
  

    

      
0
      
6.77
      
3.08
      
1
      
1
    
    

      
1
      
15.03
      
6.15
      
1
      
1
    
    

      
2
      
6.43
      
2.35
      
1
      
1
    
    

      
3
      
0.10
      
0.30
      
0
      
0
    
    

      
4
      
0.10
      
0.30
      
0
      
0
    
  

In [11]:

    

data_df.plot(kind='scatter', x='as_level', y = 'gold_proximity')


    

    
Out[11]:





<matplotlib.axes._subplots.AxesSubplot at 0x1f428ae37f0>






    

In [12]:

    

data_df.plot(kind='scatter', x='sb_level', y = 'gold_proximity')


    

    
Out[12]:





<matplotlib.axes._subplots.AxesSubplot at 0x1f428b415f8>






    

In [17]:

    

data_df['gold_proximity'].value_counts().plot(kind='bar')


    

    
Out[17]:





<matplotlib.axes._subplots.AxesSubplot at 0x1f42a0b0550>






    

In [21]:

    

counts = data_df.groupby(['gold_proximity', 'l_proximity']).size()
counts.unstack().plot(kind='bar')


    

    
Out[21]:





<matplotlib.axes._subplots.AxesSubplot at 0x1f428d18080>






    

In [24]:

    

import statsmodels.formula.api as sm
model = sm.logit(formula='gold_proximity ~ C(l_proximity)', data=data_df)
result = model.fit()
result.summary()


    

    




Optimization terminated successfully.
         Current function value: 0.360516
         Iterations 7






    
Out[24]:





Logit Regression Results

  
Dep. Variable:
  
gold_proximity
  
  No. Observations:  
  
    63
  


  
Model:
               
Logit
      
  Df Residuals:      
  
    61
  


  
Method:
               
MLE
       
  Df Model:          
  
     1
  


  
Date:
          
Mon, 17 Apr 2017
 
  Pseudo R-squ.:     
  
0.4721
  


  
Time:
              
20:00:51
     
  Log-Likelihood:    
 
 -22.712
 


  
converged:
           
True
       
  LL-Null:           
 
 -43.023
 


  
 
                      
 
        
  LLR p-value:       
 
1.848e-10




           
              
coef
     
std err
      
z
      
P>|z|
  
[0.025
    
0.975]
  


  
Intercept
           
   -2.7081
 
    0.730
 
   -3.708
 
 0.000
 
   -4.139
 
   -1.277


  
C(l_proximity)[T.1]
 
    4.1352
 
    0.860
 
    4.807
 
 0.000
 
    2.449
 
    5.821

In [45]:

    

from sklearn.model_selection import KFold
fold = KFold(n_splits=5, shuffle=True)

def cross_validate(train_index, test_index):
    train_data, test_data = data_df.iloc[train_index], data_df.iloc[test_index]
    
    t_model = sm.logit(formula='gold_proximity ~ C(l_proximity)', data=train_data)
    t_result = t_model.fit()
    t_predict = t_result.predict(test_data)
    t_predict = t_predict.apply(lambda v: 1.0 if v >= 0.5 else 0.0)
    
    return pd.concat({'predicted': t_predict, 'actual': test_data['gold_proximity']}, axis=1)

cv_results = [cross_validate(train_index, test_index) for train_index, test_index in fold.split(data_df.index)]


    

    




Optimization terminated successfully.
         Current function value: 0.414925
         Iterations 6
Optimization terminated successfully.
         Current function value: 0.277641
         Iterations 7
Optimization terminated successfully.
         Current function value: 0.337705
         Iterations 7
Optimization terminated successfully.
         Current function value: 0.408391
         Iterations 6
Optimization terminated successfully.
         Current function value: 0.335184
         Iterations 7

In [46]:

    

from sklearn.metrics import confusion_matrix
import seaborn as sns

for idx, result in enumerate(cv_results):
    c = confusion_matrix(result['predicted'], result['actual'])
    pylab.figure()
    ax = sns.heatmap(c, annot=True)
    ax.set_title('Confusion Matrix for fold %s' % (idx+1))


    

In [48]:

    

from sklearn.metrics import f1_score

scores = [f1_score(r['predicted'], r['actual']) for r in cv_results]
sc = pd.Series(data=scores, name='F1_scores')
sc.plot(kind='bar', title='5-fold cross validation results')


    

    
Out[48]:





<matplotlib.axes._subplots.AxesSubplot at 0x1f42c5c2908>