Import the package



In [1]:

    
from grmpy.test.auxiliary import refactor_results
from scipy.stats import gaussian_kde
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
%matplotlib inline  

import grmpy









    



/home/sebastian/anaconda3/envs/grmpy_tutorial/lib/python3.6/site-packages/statsmodels/compat/pandas.py:56: FutureWarning: The pandas.core.datetools module is deprecated and will be removed in a future version. Please use the pandas.tseries module instead.
  from pandas.core import datetools

Simulation Process



In [31]:

    
df = grmpy.simulate('tutorial.grmpy.ini')



In [4]:

    
df.head()

The simulation process provides a the simulated data as a pickle file



In [6]:

    
pickle_file = pd.read_pickle('data.grmpy.pkl')
pickle_file.head()

Plot - Treatment effects and benefit distribution



In [7]:

    
benefits= df.Y1 - df.Y0
TT = np.mean(benefits[df.D==1])
TUT = np.mean(benefits[df.D==0])
ATE = np.mean(benefits)
density = gaussian_kde(benefits)

xs = np.linspace(-4,5)
plt.figure(figsize=(20,10)) 

plt.plot(xs,density(xs))
plt.axvline(x=TUT, c='g', label='TUT')
plt.axvline(x=TT, c='r', label='TT')
plt.axvline(x=ATE, c='y',label='ATE')
plt.xlabel('Y1-Y0', fontsize=20)
plt.ylabel('Density', fontsize=20)
plt.legend(fontsize=20)

plt.show()



In [28]:

    
df_ = grmpy.simulate('tutorial2.grmpy.ini')



In [29]:

    
benefits_ = df_.Y1 - df_.Y0
TT = np.mean(benefits_[df.D==1])
TUT = np.mean(benefits_[df.D==0])
ATE = np.mean(benefits_)
density = gaussian_kde(benefits_)

xs = np.linspace(-4,5)
plt.figure(figsize=(20,10)) 

plt.plot(xs,density(xs))
plt.axvline(x=TUT, c='g', label='TUT')
plt.axvline(x=TT, c='r', label='TT')
plt.axvline(x=ATE, c='y',label='ATE')
plt.xlabel('Y1-Y0', fontsize=20)
plt.ylabel('Density', fontsize=20)
plt.legend(fontsize=20)

plt.show()

Estimation

init: Init file values as start values



In [41]:

    
results = grmpy.estimate('tutorial.grmpy.ini')



In [42]:

    
refactor_results(results, 'tutorial.grmpy.ini')

Plot Estimation Results



In [43]:

    
df2 = grmpy.simulate('test.grmpy.ini')
benefits_est = df2.Y1 - df.Y0
density = gaussian_kde(benefits)
density_est = gaussian_kde(benefits_est)
xs = np.linspace(-4,5)
plt.figure(figsize=(20,10)) 

plt.plot(xs, density_est(xs), c='g', label='est')
plt.plot(xs, density(xs), c='b', label='sim')
plt.xlabel('Y1-Y0', fontsize=20)
plt.ylabel('Density', fontsize=20)
plt.legend(fontsize=20)


plt.show()

auto: Start values are derived from a simple OLS and a Probit regression



In [44]:

    
results = grmpy.estimate('tutorial_auto.grmpy.ini')



In [45]:

    
refactor_results(results, 'tutorial_auto.grmpy.ini')

Plot Estimation Results



In [46]:

    
df2 = grmpy.simulate('test.grmpy.ini')
benefits_est = df2.Y1 - df.Y0
density = gaussian_kde(benefits)
density_est = gaussian_kde(benefits_est)
xs = np.linspace(-4,5)
plt.figure(figsize=(20,10)) 

plt.plot(xs, density_est(xs), c='g', label='est')
plt.plot(xs, density(xs), c='b', label='sim')
plt.xlabel('Y1-Y0', fontsize=20)
plt.ylabel('Density', fontsize=20)
plt.legend(fontsize=20)


plt.show()



In [ ]:

	Y	X_0	X_1	X_2	X_3	Z_0	Z_1	Y1	Y0	U1	U0	UC	V
0	3.682474	1.0	0.0	1.012175	-0.653786	1.0	0.514857	4.704468	3.682474	0.367474	0.410380	0.484450	0.527356
1	1.558495	1.0	1.0	0.682808	-0.590194	1.0	0.609202	5.647489	1.558495	0.847235	-1.789534	4.309898	1.673129
2	0.241303	1.0	0.0	-0.036321	-0.196745	1.0	-0.495743	2.341456	0.241303	0.095008	-1.420377	2.287648	0.772262
3	1.981197	1.0	1.0	-0.534595	-0.499604	1.0	0.141740	2.781138	1.981197	-0.251911	0.052811	-0.438839	-0.134117
4	1.235780	1.0	1.0	-0.850554	-0.201994	1.0	0.603826	-0.617849	1.235780	-2.781097	-0.043756	-2.205267	0.532074

	Y	X_0	X_1	X_2	X_3	Z_0	Z_1	Y1	Y0	U1	U0	UC	V
0	3.682474	1.0	0.0	1.012175	-0.653786	1.0	0.514857	4.704468	3.682474	0.367474	0.410380	0.484450	0.527356
1	1.558495	1.0	1.0	0.682808	-0.590194	1.0	0.609202	5.647489	1.558495	0.847235	-1.789534	4.309898	1.673129
2	0.241303	1.0	0.0	-0.036321	-0.196745	1.0	-0.495743	2.341456	0.241303	0.095008	-1.420377	2.287648	0.772262
3	1.981197	1.0	1.0	-0.534595	-0.499604	1.0	0.141740	2.781138	1.981197	-0.251911	0.052811	-0.438839	-0.134117
4	1.235780	1.0	1.0	-0.850554	-0.201994	1.0	0.603826	-0.617849	1.235780	-2.781097	-0.043756	-2.205267	0.532074