A High-Dimensional Data Regression Model Based on Stacking-Gaussian Process Algorithm

Abstract

In order to mine valuable information and potential rules from a large amount of data, processing high-dimensional data and establishing regression prediction models has become a hot research topic in the field of machine learning. High-dimensional input and nonlinear relationship fitting are two common problems in establishing regression models. Based on this, this paper proposes a high-dimensional data regression model based on a stacking framework and Gaussian process algorithm. Specifically, on the one hand, the stacking framework is used to alleviate the curse of dimensionality problem caused by high input dimensions. On the other hand, based on different kernel functions, the Gaussian process regression model can fit linear or nonlinear connection functions of different complexity. Simulation experiments and real data analysis show that the proposed method has smaller prediction errors, better prediction accuracy, and good generalization performance compared to some contrastive methods such as Random Forests, Lasso Regression, and Back Propagation Neural Network, demonstrating the superiority of the proposed model.