User Behavior Prediction Model Based on Deep Learning Algorithms in the Digital Economy Environment

Economic globalization has driven the rapid development of the digital economy, with cross-border e-commerce rising alongside it. In this context, the accurate prediction of user behavior is crucial for shaping future marketing strategies and optimizing the allocation of social resources. To address this, this paper proposes a user behavior prediction model based on deep learning algorithms. The model integrates the strengths of three fundamental models: Logistic Regression, Support Vector Machine (SVM), and Random Forest, enhancing the overall performance. In cases where the feature matrix is extremely sparse, the model efficiently predicts user behavior by utilizing second-order feature interactions. Experimental results indicate that the importance of the predictive factors for user behavior is as follows: Product Functionality (D) > Influencer Recommendations (B) > User Inertia (E) > Product Quality (C) > Brand Image (A). Additionally, the model's normalized Gini coefficient reaches a maximum value of 0.77, with a prediction accuracy of 0.95 and a recall rate of 0.94 in both iterations, demonstrating excellent predictive performance. Overall, the deep learning algorithm-based prediction model outperforms all other comparison models in terms of accuracy, recall, stability, and generalization capability, making it the most effective model among those tested.