Meta - Heuristic Driven Loss Ratio Optimization

The material described in this thesis is the result of an internship of the master study Business Analytics. It has been conducted at FRISS from April 5, 2015 until December 4, 2015. FRISS is a consultancy and software company specialized in fraudand risk management for the insurance sector and is located in Utrecht, the Netherlands. The main research problem described in this thesis is how to find high quality solutions to constrained Portfolio Value Optimization problems by using modern metaheuristic search procedures. A number of distinct search procedures were implemented and extensively tested. The experiments described in this work indicate that in practice high quality solutions can be found in short timespans and under a wide range of conditions.