(Data processing done in ArcGIS outside of this repository.)
Using TxDOT's roadway inventory and CRIS crash databases, predict crash rates from roadway infrastructure features using a random forest model and compare against the best practice FHWA safety performance functions model (latter of which is also done externally).
Neural net model predicting crash rates without restricting inputs to just infrastructure roadway features.