Python package: ATE

The main aim of ATE is to provide a user-friendly interface for nonparametric efficient inference of average treatment effects for observational data. The package provides point estimates for average treatment effects, average treatment effect on the treated and can also handle the case of multiple treatments. The package also allows inference by consistent variance estimates.

Requirements

• numpy
• pandas
• scipy
• matplotlib

Installation

The package can be installed from pypi:

pip install ATE

Alternatively, we can directly install from Github:

pip install git+https://github.com/ross1456/ATE-python.git

Key Features

• Ease of use: The main function ATE requires only a numeric matrix X of covariates, numeric vector Y of response and treat vector indicating treatment assignment.

#Generate some data
import numpy as np
np.random.seed(0)
n = 500
# Generate covariates
X1 = np.random.normal(size=(n, 5))
X2 = np.random.binomial(1, 0.4, size=(n, 3))
X = np.hstack((X1, X2))

# Calculate propensity scores
prop = 1 / (1 + np.exp(X[:, 0] - 0.5 * X[:, 1] + 0.25 * X[:, 2] + X[:, 5] + 0.5 * X[:, 7]))

# Treatment assignment
treat = np.random.binomial(1, prop, size=n)

# Outcome variable
Y = 10 * treat + (2 * treat - 1) * (X[:, 0] - 0.5 * X[:, 1] + 0.25 * X[:, 2] + X[:, 5] + 0.5 * X[:, 7]) + np.random.normal(size=n)


#Fit ATE object
ate1 <- ATE()
ate1.fit(Y, treat, X)
ate1.summary()

          Estimate  Std. Error  95%.Lower  95%.Upper    z value       p value
E[Y(1)]  10.522690    0.106475  10.314003  10.731376  98.827916  0.000000e+00
E[Y(0)]  -0.640634    0.080557  -0.798523  -0.482746  -7.952586  1.826574e-15
ATE      11.163324    0.157352  10.854920  11.471728  70.944896  0.000000e+00

• We can also estimate the average treatment effect on the treated.

ate2 <- ATE(ATT = TRUE)
ate2.fit(Y, treat,X)
ate2.summary()

         Estimate  Std. Error  95%.Lower  95%.Upper    z value   p value
E[Y(1)]  9.702759    0.100730   9.505331   9.900187  96.324138  0.000000
E[Y(0)]  0.265824    0.121707   0.027283   0.504365   2.184136  0.028952
ATE      9.436935    0.188530   9.067422   9.806448  50.055253  0.000000

• ATE automatically detects and estimates the case of multiple treatment arm.

ate3<-ATE()
ate3.fit(Y,treat,X)
ate3.summary()

          Estimate  Std. Error  95%.Lower  95%.Upper   z value       p value
E[Y(0)]  -0.433280    0.332050  -1.084086   0.217526 -1.304863  1.919395e-01
E[Y(1)]  10.391241    1.421770   7.604622  13.177860  7.308663  2.698139e-13
E[Y(2)]  21.393437    5.246493  11.110499  31.676375  4.077664  4.549043e-05
E[Y(3)]  32.643964   12.081066   8.965510  56.322417  2.702077  6.890790e-03