This repository stores the code-base for my CS221 (Introduction to Artificial Intelligence) final project on generating a predictive model for poverty from pollution metrics.

Environmental discrimination is the disproportionate exposure of low income areas and minority groups to the negative impacts of pollution. Environmental organizations consider it an injustice that those mired in poverty suffer the most from the impacts of pollution despite contributing the least. In our project, we determined the extent to which poverty can be predicted by the levels of air pollution. To do so, we scraped EPA (Environmental Protection Agency) air pollution data by county from 2000 to 2017 and the corresponding household income data from the U.S. Census Bureau. As our baseline, we ran a linear regression on our data-set. Then we implemented two models, a logistic regression and multi-layered linear regression neural network, as predictors of the median income bracket. Both models clearly outperformed the baseline.