Geospatial Prediction of Soil Erosion in Cameron Highlands Using Decision Tree Technique

Soil erosion issue in Cameron Highland has caused landslide, mudflow, river bank erosion and contamination of drinking water. This study is focused on predicting the location of soil erosion in the study area using decision tree technique. The selected causative factors are length slope (LS), normalized difference vegetation index (NDVI), land surface temperature (LST), erosivity (R), drainage density (DD) and lineament density (LD). Based on selected causative factors, maps of CF’s were produced by using ArcGIS software. Data extracted from all six the causative factors mapping were trained and tested in WEKA. In WEKA, there are many types of decision trees that could be used to analyze data. Among the type of decision trees available are decision stump, Hoeffding tree, J48, LMT, M5P, random forest, random tree and REP tree. For this study, the result was obtained by using the random forest method. The outcome of this study shows good result. The correlation coefficient in the result summary table indicates the reliability of the run analysis. It indicates the explanatory power of the analysis. As for the training and testing result, such values of coefficient of correlation show successful run.