In [20]:

    

%matplotlib inline


    

In [21]:

    

import statsmodels.api as sm
import statsmodels.formula.api as smf
import pandas as pd
import seaborn as sns
import matplotlib as mpl
import matplotlib.pyplot as plt


    

In [22]:

    

bodyfat = pd.read_fwf('bodyfat.txt', names=['dietfat', 'bodyfat'])

fatmodel = smf.ols('bodyfat ~ dietfat', bodyfat)

fatresults = fatmodel.fit()


    

In [34]:

    

fatresults.summary()


    

    
Out[34]:





OLS Regression Results

  
Dep. Variable:
         
bodyfat
     
  R-squared:         
 
   0.376


  
Model:
                   
OLS
       
  Adj. R-squared:    
 
   0.337


  
Method:
             
Least Squares
  
  F-statistic:       
 
   9.659


  
Date:
             
Tue, 06 Jan 2015
 
  Prob (F-statistic):
  
0.00677


  
Time:
                 
18:01:28
     
  Log-Likelihood:    
 
 -27.595


  
No. Observations:
      
    18
      
  AIC:               
 
   59.19


  
Df Residuals:
          
    16
      
  BIC:               
 
   60.97


  
Df Model:
              
     1
      
                     
     
 
   




      
         
coef
     
std err
      
t
      
P>|t|
 
[95.0% Conf. Int.]
 


  
Intercept
 
    7.3141
 
    1.006
 
    7.271
 
 0.000
 
    5.182     9.447


  
dietfat
   
    0.1162
 
    0.037
 
    3.108
 
 0.007
 
    0.037     0.196




  
Omnibus:
       
 0.259
 
  Durbin-Watson:     
 
   2.156


  
Prob(Omnibus):
 
 0.879
 
  Jarque-Bera (JB):  
 
   0.008


  
Skew:
          
-0.002
 
  Prob(JB):          
 
   0.996


  
Kurtosis:
      
 2.894
 
  Cond. No.          
 
    96.7

In [23]:

    

sm.qqplot(fatresults.resid, line='s')


    

    
Out[23]:












    

In [42]:

    

plt.plot(fatresults.resid, 'o')


    

    
Out[42]:





[<matplotlib.lines.Line2D at 0x10e44b198>]






    

In [25]:

    

sns.lmplot('dietfat', 'bodyfat', bodyfat)


    

    
Out[25]:





<seaborn.axisgrid.FacetGrid at 0x10d6c7898>






    

In [29]:

    

# 2nd order polynomial
sns.lmplot('dietfat', 'bodyfat', bodyfat, order=2)


    

    
Out[29]:





<seaborn.axisgrid.FacetGrid at 0x10da82d68>






    

In [31]:

    

# Residual plot
sns.residplot('dietfat', 'bodyfat', bodyfat)


    

    
Out[31]:





<matplotlib.axes._subplots.AxesSubplot at 0x10dbab208>






    

In [37]:

    

electric = pd.read_fwf('electric.txt', names=['bill', 'income', 'people', 'sqft'])

emodel = smf.ols('bill ~ income + people + sqft', electric)

eresults = emodel.fit()


    

In [38]:

    

eresults.summary()


    

    
Out[38]:





OLS Regression Results

  
Dep. Variable:
          
bill
       
  R-squared:         
 
   0.840


  
Model:
                   
OLS
       
  Adj. R-squared:    
 
   0.824


  
Method:
             
Least Squares
  
  F-statistic:       
 
   52.36


  
Date:
             
Tue, 06 Jan 2015
 
  Prob (F-statistic):
 
4.92e-12


  
Time:
                 
18:21:59
     
  Log-Likelihood:    
 
 -214.29


  
No. Observations:
      
    34
      
  AIC:               
 
   436.6


  
Df Residuals:
          
    30
      
  BIC:               
 
   442.7


  
Df Model:
              
     3
      
                     
     
 
   




      
         
coef
     
std err
      
t
      
P>|t|
 
[95.0% Conf. Int.]
 


  
Intercept
 
 -554.9445
 
   98.575
 
   -5.630
 
 0.000
 
 -756.261  -353.628


  
income
    
    0.2476
 
    0.024
 
   10.531
 
 0.000
 
    0.200     0.296


  
people
    
   82.1051
 
   16.283
 
    5.042
 
 0.000
 
   48.851   115.359


  
sqft
      
   -0.0144
 
    0.014
 
   -1.027
 
 0.313
 
   -0.043     0.014




  
Omnibus:
       
 1.293
 
  Durbin-Watson:     
 
   2.134


  
Prob(Omnibus):
 
 0.524
 
  Jarque-Bera (JB):  
 
   1.173


  
Skew:
          
 0.423
 
  Prob(JB):          
 
   0.556


  
Kurtosis:
      
 2.665
 
  Cond. No.          
 
1.97e+04

In [39]:

    

sns.corrplot(electric)


    

    
Out[39]:





<matplotlib.axes._subplots.AxesSubplot at 0x10d877da0>






    

In [ ]: