Using a dataset of remote sensed features of an air production unit on a metro transit train to predict when measurements from sensors are indicative of being within a week of an air leak failure due to high stress.
This notebook includes data preprocessing, data analysis, model training/evaluating/validating/optimizing, and identification of features most predictive of a an impending failure.