Loading content...
The Genetic Algorithm (GA) is applied simultaneously for optimal feature subset selection and Random Forest parameter tuning. Each chromosome consists of a binary feature mask and a set of continuous hyperparameters. The fitness function is the AUPRC, which provides an assessment of classification efficiency and also penalises large feature subsets to encourage compactness. The Genetic Algorithm operations where the selection crossover and mutation of DNA are shown in Algorithm 2, iteratively evolve this population supporting gradual refinement towards optimal configurations. This evolutionary method allows effective exploration of a large search space to achieve performance trade-offs between accuracy, computational resources and interpretability in fraud detection tasks.
The Genetic Algorithm operations where the selection crossover and mutation of DNA are shown in Algorithm 2, iteratively evolve this population supporting gradual refinement towards optimal configurations. This evolutionary method allows effective exploration of a large search space to achieve performance trade-offs between accuracy, computational resources and interpretability in fraud detection tasks.