Shapley Feature Selection

Alex Gramegna, Paolo Giudici
2022 FinTech  
Feature selection is a popular topic. The main approaches to deal with it fall into the three main categories of filters, wrappers and embedded methods. Advancement in algorithms, though proving fruitful, may be not enough. We propose to integrate an explainable AI approach, based on Shapley values, to provide more accurate information for feature selection. We test our proposal in a real setting, which concerns the prediction of the probability of default of Small and Medium Enterprises. Our
more » ... sults show that the integrated approach may indeed prove fruitful to some feature selection methods, in particular more parsimonious ones like LASSO. In general the combination of approaches seems to provide useful information which feature selection algorithm can improve their performance with.
doi:10.3390/fintech1010006 fatcat:kie7gvgmajcctmtb7vnmq2lhwy