In this work, we explore the possibility of employing evolutionary algorithms to choose the best set of data points that meet a particular criterion. The sampling is applied to data from both sources (simulations and real experiments). A well-dispersed data point will result from such an action, reducing the chance of producing clustered data while merging both datasets. The established approach is applied to a manufacturing case study, in which two data sources need to be merged in order to improve the prediction of product quality.
The current strategy is based on the use of a global direct optimiser or explorer, such as a genetic algorithm.