Dragonnet reproduced

This repository contains software and data reproducing the dragonnet implementation by "Shi et al." This is done through rewriting certain parts of their implementation with PyTorch. The paper that belongs to this repository is "Adapting Neural Networks for the Estimation of Treatment Effects". The paper describes the use of neural networks for the estimation of treatment effects from observational data.

Requirements and setup

You will need to install sklearn, numpy, pytorch and, pandas

Data

1. IHDP

• This dataset is based on a randomized experiment investigating the effect of home visits by specialists on future cognitive scores.
• It is generated via the npci package.
• We also uploaded a portion of the simulated data in the dat folder.

2. ACIC

• ACIC is a collection of semi-synthetic datasets derived from the linked birth and infant death data (LBIDD) Here is the full dataset description data set description

• Here is the GitHub repo associated with the competition IBM benchmark For access to the ACIC 2018 competition data: Please see here [ACIC data] (https://www.synapse.org/#!Synapse:syn11294478/wiki/486304)

Reproducing the IHDP experiments

The workflow consists of two stages:

1. Fitting a predictor:

• You'll run the from src code as ./experiment/run_ihdp.sh

• If you are using a cluster, there's some code that might be useful in the submission folder.

• Before doing this, you'll need to edit run_ihdp.sh and change the following: data_base_dir= where you stored the data output_base_dir=wherer you want the result to be

• The prediction should go into your output folder

2. Estimation:

• After you get the predictions, you want to fit them into estimators.
• Run ihdp_ate.py to reproduce the table.

3. Things to note when reproducing the result:

• The default setting would let you run Dragonnet, and TARNET under both targeted regularization and default mode. If you want to run a subset of the models, delete them at the run_ihdp.sh

• The default use all the data for training, prediction, and estimation. To change that, you could update the train_test_split function in the ihdp_main.py

Reproducing the ACIC experiment

1. Fitting a predictor: Same as above except you run the from src code as ./experiment/run_acic.sh

2. Estimation: Same as above except you run the from src code as acic_ate.py

Contact us

This repository is created by Alec Nonnemaker, Sara Boby and Fernando Corte Vargas. Feel free to email us if you have any questions: [email protected]