Cross-Border E-Commerce Consumer Behavior Big Data Analysis and Prediction Model Research

This paper introduces a hybrid model, G-GBDBN, which integrates Gaussian Process Regression (GPR) with an enhanced Deep Belief Network (GrDBN). The model introduces a continuous restricted Boltzmann machine (RBM) in the DBN framework to solve problems such as data loss in an unsupervised mode. In addition, it uses a Gaussian activation function to enhance the generalization ability and stability of the model. Through this innovative combination, the G-GBDBN model can more effectively capture the nonlinear features of data and improve prediction accuracy. Experimental results show that the G-GBDBN model outperforms traditional regression models and deep learning models on multiple datasets, especially in handling high-dimensional, nonlinear data, demonstrating stronger fitting ability and higher prediction accuracy. This model provides an effective solution for solving complex prediction problems and is especially suitable for big data analysis and pattern recognition in various fields.