Fast Classification Model Based on Genetic Algorithm and XGBoost-RandomForest Stacking Model

周彦良

Abstract: The balance between calculation accuracy and running time is a problem in effectively utilizing the stacking model. In this paper, a hybrid ensemble learning model is introduced to improve computational accuracy, and genetic algorithm is used to select features in the training process to reduce running time. The model is built by testing a single model, selecting a model with good performance to form a stack model, and then using genetic algorithm to select features on the stacking model. The proposed model demonstrates superior comprehensive performance compared to both individual models and stacking models, as verified by analysis of three datasets.

Submission Number: 2