A Model Selection Method for Machine Learning by Differential Evolution

While the application of big data attracts more and more attention, machine learning algorithms are developing with each passing day, and the models produced by machine learning are increasingly diversified. The focus of big data applications has gradually shifted from model training to the prediction and inference. Choosing the most suitable model for enterprise application scenarios among many machine learning models has become a topic of research that has attracted much attention. Though ensemble methods have been proposed to discover best model by multiple training phase, studies of finding best combination within multiple modes are still few. Configuring different machine learning models with appropriate parameters and looking for parameters is an NP-hard problem, which requires an optimization algorithm. This study proposes a differential evolution algorithm to integrate multiple trained machine learning models into an appropriate model. In this paper, the regression model is taken as an example and the differential evolution algorithm is compared with the genetic algorithm. Three benchmark datasets are used to examine, and the results show that the differential evolution algorithm outperforms genetic algorithm.