A genetic algorithm approach to the selection of near-optimal subsets from large sets

The problem attempted in this paper is to select a sample from a large set where the sample is required to have a particular average property. The problem can be expressed as an optimisation problem where one selects a subset of r objects from a group of n objects and the objective function is the mismatch between the required average property and that of a proposed sample. We test our method on a reallife problem which arises when we model the assets of a life insurance company in order to understand its risk, solvency and/or capital requirements. In this paper we describe a genetic algorithm developed to solve the generic selection task. We demonstrate the algorithm successfully solving our test problem. Categories and Subject Descriptors I.2.8 [Computing Methodologies]: Artifical Intelligence— Heuristic Methods General Terms Algorithms Keywords Genetic Algorithm, Sampling, Selection, Economics

• Type: Conference
• Year: 2005
• Where: GECCO
• Authors: P. Whiting, P. W. Poon, J. N. Carter
• Keywords: Average Property, GECCO 2005, Genetic Algorithm, Particular Average Property