Probability Estimation by an Adapted Genetic Algorithm in Web Insurance

• Published in: Learning and Intelligent Optimization Vol. 11353; pp. 225 - 240
• Main Authors: Bedenel, Anne-Lise, Jourdan, Laetitia, Biernacki, Christophe
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 01.01.2019; Springer International Publishing
• Series: Lecture Notes in Computer Science
• Subjects: BIC; Genetic Algorithms; Insurance; WEB
• ISBN: 3030053474; 9783030053475
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-030-05348-2_21

In the insurance comparison domain, data constantly evolve, implying some difficulties to directly exploit them. Indeed, most of the classical learning methods require data descriptors equal to both learning and test samples. To answer business expectations, online forms where data come from are regularly modified. This constant modification of features and data descriptors makes statistical analysis more complex. A first work with statistical methods has been realized. This method relies on likelihood and models selection with the Bayesian information criterion. Unfortunately, this method is very expensive in computation time. Moreover, with this method, all models should be exhaustively compared, what is materially unattainable, so the search space is limited to a specific models family. In this work, we propose to use a genetic algorithm (GA) specifically adapted to overcome the statistical method defaults and shows its performances on real datasets provided by the company MeilleureAssurance.com.