Markov modelling and parameterisation of genetic evolutionary test generations

Abstract

Genetic evolutionary algorithms are effective and optimal test generation methods. However, the methods to select the algorithm parameters are often ad hoc, relying on empirical data. We used a Markov-based method to model the genetic evolutionary test generation process, parameterise the process characteristics, and derive analytical solutions for selecting the optimisation parameters. The method eliminates preliminary test generation calibration and experimentation effort needed to select these parameters, which are used in current practice.