Flight Delay Prediction

Flight delays are 33B addressable problem and the goal of the project was to predict if a flight will be delayed two hours before the departure time. From the data, we could see that we were dealing with an imbalanced classification problem and one class,in our case,on time flights represent majority of the data points.  Metrics like recall and precision are good metrics to understand the usefulness of our models. We focused on optimizing the recall for our models,while also making sure precision does not suffer. The outcome was an ensemble of multiple ML models with majority voting approach.