Poverty Prediction by Satellite Images and Deep Learning

Chiyuan Cheng (08/2020)

This project uses transfer learning to predict poverty (Wealth index) of a sub-Saharan African country, Burundi, in 2010.
Regression models are used to predict Wealth index from luminosity of nighttime satellite images, with r-squared of 0.54.
Gaussian Mixture Model is used to classify the daytime satellite imagery into three classes, based on the luminosity.
Transfer learning (VGG16, ResNet50, Inception C3) to capture features from daytime satellite imagery and predict poverty, with the 80% accuracy from the best model.

Name		Name	Last commit message	Last commit date
Latest commit History 27 Commits
.ipynb_checkpoints		.ipynb_checkpoints
figures		figures
01_DHS_prep.ipynb		01_DHS_prep.ipynb
02_Satellite_Image_Prep.ipynb		02_Satellite_Image_Prep.ipynb
03_Satellite_Image_nighttime_ML.ipynb		03_Satellite_Image_nighttime_ML.ipynb
04_Satellite_Image_process.ipynb		04_Satellite_Image_process.ipynb
05_Satellite_Image_similarity.ipynb		05_Satellite_Image_similarity.ipynb
06_Satellite_Image_TL.ipynb		06_Satellite_Image_TL.ipynb
Capstone Project 2_ Final Report.pdf		Capstone Project 2_ Final Report.pdf
Capstone2_presentation.pdf		Capstone2_presentation.pdf
DHS_light.csv		DHS_light.csv
README.md		README.md

Provide feedback