Parameter importance analysis: Random forest approach

Abstract During surface roughness modelling, it is crucial to determine the parameters with highest predictive power since these are outcome drivers. Based on out-of-bag (OOB) mean square error, following Random Forest techniques have been utilized parameter importance: decrease in accuracy and total increase node purity. Validation of results has achieved using Bayesian linear regression technique. The PMMA machining experiment designed by Central Composite Design (CCD) Face Centered Cutting speed, feed rate depth cut control parameters, while quality dependent parameter. authors established that random forest algorithm yields an OOB squared error 0.113 decreases increasing number trees for validation dataset. On other hand, increases training Both purity reveal order decreasing importance as follows: cutting rate. obtained same outcome. Hence, may be omitted from models faster simpler prediction.