This document summarizes a research project that aims to predict flight delays using machine learning models. The author analyzes flight delay data from the Bureau of Transportation Statistics to build a random forest model that predicts whether a flight will be delayed or not based on features like date, airline, and origin/destination airports. After preprocessing the data, the author fits a random forest model and evaluates its accuracy at 67%. Some limitations of the current approach are discussed as well as lessons learned regarding validating multiple models and potentially important additional predictor variables.