Datasets:
dataset:
- name: >-
Replication Data for Integrating Earth Observation Data into Causal
Inference
tags:
- causal-inference
- earth-observation
- satellite-imagery
- geospatial
- observational-data
- research-data
license: mit
Replication Data for: Integrating Earth Observation Data into Causal Inference: Challenges and Opportunities
Details:
YandW_mat.csv
contains individual-level observational data. In the dataset, LONGITUDE
and LATITUDE
refer to the approximate geo-referenced long/lat of observational units. Experimental outcomes are stored in Yobs
. The treatment variable is stored in Wobs
. The unique image key for each observational unit is saved in UNIQUE_ID
.
Geo-referenced satellite images are saved in
./Nigeria2000_processed/%s_BAND%s.csv
, where the first "%s
" refers to the image key associated with each observation (saved in UNIQUE_ID
in YandW_mat.csv
) and BAND%s
refers to one of 3 bands in the satellite imagery.
Code Link: https://github.com/cjerzak/causalimages-software/
Paper Reference: Connor T. Jerzak, Fredrik Johansson, Adel Daoud. Integrating Earth Observation Data into Causal Inference: Challenges and Opportunities. ArXiv Preprint, 2023. [PDF]