FIGURE 2.

Overview of the models count for each generation from one repetition for the different configurations experiments. Models with a darker colour were more popular then models with lighter colour. Across the four experiments, random forest, K‐nearest neighbours, linear regression and extra trees regressors are the models with the highest count per generation. To make sure that all models were represented, we had 1,000 models in the first generation and 100 models were passed on for the following generations