Python codes for real-time maximum flood inundation modelling using machine learning

This methodology involves developing two random forest (RF) models, one for predicting maximum depth (RFD) and another model for predicting maximum velocity (RFV).

The codes contain two identical folders with the name RFD folder contains all the python codes for predicting maximum depth and RFV folder contains python codes for predicting maximum velocity.

Each folder contains two sub folders, namely, featureSelection and hyperParameterization. The “featureSelection” folder contains the feed forward method codes used for selecting the best physical characteristics. “hyperParameterization” folder contains two sub folders, namely, “maxFeatures” and “n_tress”. These folders contains the corresponding codes. If the file name contains “train” at the end, which means it is for training purposes and if it contains “test” then it is used for testing purposes.

“RF_Final_maxDepth” and “RF_Final_maxVel” are the final trained files used for predicting maximum depth and velocity on testing and validations events, respectively.

The “RFD” folder contains an extra folder named, “improvedRFD”. This folder contains all the python files for improving the RFD model. The file name includes the combination name as “c*”, where “*” is the combination number and “train” and “test” keywords in the file name indicate the training and testing files.