In this demo, Classification And Regression Tree (CART), Gradient Boosting Machine (GBM), & Support Vector Machine (SVM) are all applied in R to create flood susceptibility maps of the Red River Valley in Manitoba, Canada. Each demo runs through the traditional methods of creating a susceptibility map, including separating out training and testing data, running a Pearson’s coefficient matrix, and then analysis of model performance through ROC curve plotting and Kappa index scoring. More detail on the demo methods are included in the "Final_Report" word document. The sample data was included in this repository (Red_River_Flood_Samples.csv), however the full dataset is available HERE
This demo was made to test how R could be used to create flood susceptibiltiy maps, but the tecchniques used here could be applied to other types of susceptibility mapping(Landslide, Fire, etc.). Some sources for the code are in the script comments, and the final report includes citations to scientific journals and sources for the data used in this demo. Please let me know what you think of this demo, and if you want to use it for a class or a tutorial, please email me at blair.scriven@ucalgary.ca