Application of Recursive Feature Elimination (RFE) in Profiling Contracting Performance of Contractors in the Province of South Cotabato

Profiling contracting performance involves heavy obligations. The need to consider various factors and many decision parameters for an effective assessment of the contractor is considered a multi-criteria problem which made the contract selection onerous. In the process of Contractor selection, this study proposes an alternative way in profiling contractor performance in the province of South Cotabato using Recursive Feature Elimination (RFE). RFE (Recursive Feature Elimination) is a type of feature selection method that:

a. Performs classifier training to find the weight vector,

b. Calculate ranking criteria for all features,

c. Eliminate features with the smallest ranking criterion value.

Several data sets have been checked and tested applying the Recursive Feature Elimination (RFE). The experiment shows Recursive Feature Elimination with Gradient Boosting performs best in a dataset with many features identifying minimal numbers of features with high accuracy and is effective in reducing the number of selected features.