emilylaiken/togo-targeting-replication

# Togo Targeting Replication Code ## Introduction This repository contains replication code for the paper "Machine Learning and Mobile Phone Data Can Improve Targeting of Humanitarian Assistance" by Emily Aiken, Suzanne Bellue, Dean Karlan, Chris Udry, and Joshua Blumenstock. This readme provides information about the code structure, including where replication code for each figure and table in the paper is located, and information about the data available. ## Code Structure All code is written in Jupyter Notebooks and raw Python, using Python version 3.6. Replication code for all figures and tables that are generated with code is provided. It is assumed that the reader already has access to featurized mobile phone data matched to survey observations (as shown in the synthetic data provided, see section on "data structure" for more information). The scripts are organized into five notebooks as follows. All notebooks call helper functions from the file `helpers.py`. #### `1survey.ipynb` Code for all figures and tables that are generated from mainly from survey data, including the calculation of the PMT and asset index, statistics on phone ownership, summary statistics from surveys, and information on weighting and response weights. Replication code for the following tables and figures is included in this notebook: - Supplementary Figure 1 - Supplementary Figure 2 - Supplementary Figure 3 - Supplementary Figure 4 - Supplementary Figure 10 - Supplementary Figure 11 - Supplementary Table 2 - Supplementary Table 3 - Supplementary Table 4 - Supplementary Table 5 - Supplementary Table 6 - Supplementary Table 10 #### `2satellite.ipynb` Code for aggregation of satellite-based wealth estimates and satellite-based population density estimates for high-resolution poverty mapping. Replication code for the following tables and figures is included in this notebook: - Extended Data Figure 1 Panel A - Supplementary Figure 5 #### `3ml.ipynb` Code for machine learning from featur …