Eric LecoutreVan Oirbeek, RobinKounima, OthmaneOthmaneKounima2025-05-142025-05-142025-05-142023https://hdl.handle.net/2078.2/34677This master thesis is an exploration of both Information Value (IV) and Weight-of-Evidence (WoE) measures as univariate variable selection methods, going through their computation and specific needs when working with a binary target variable, and attempting to adapt them to a continuous target, which is still a barely explored subject. For this purpose, both measures are studied thoroughly as well as multiple binning approaches for the importance it holds in computing these measures. Finally, An adapted IV formula is proposed and applied on a real insurance data set in order to select an optimal set of main and interaction terms, and to compare it to other feature selection methodsInformation ValueWeight-of-EvidenceUnivariate feature selectionBinningtext::thesis::master thesisthesis:43399