Scholarlink

In the e-commerce domain, accurate prediction of consumer behavior is crucial for optimizing marketing strategies. This study conducts a predictive analysis of consumer purchase intention using the UCL e-commerce platform dataset, employing systematic data preprocessing, feature engineering, and multiple machine learning models (including Support Vector Machine, K-Nearest Neighbors, and Decision Tree). Furthermore, we propose a hybrid approach that integrates deep learning techniques with the SVM model. Through comprehensive experiments, we demonstrate that high-dimensional data and nonlinear feature engineering significantly enhance prediction accuracy. Correlation analysis reveals “PageValues” and “ExitRates” as the most influential features affecting purchase decisions. Experimental results show that the proposed Deep Neural Network-enhanced SVM model achieves the highest prediction accuracy of 91.14%, outperforming baseline models. This research provides scientific guidance for e-commerce platforms to optimize marketing strategies and offers novel methodological insights for digital consumer behavior analysis.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Digital Consumer Purchase Intention Prediction Using DNN-Improved SVM and Machine Learning Models

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Yinwen Wang,
  • Yi Huang,
  • Ling Peng,
  • Yiwen Liu,
  • Haoyu Lei,
  • Yujun Yang

摘要

In the e-commerce domain, accurate prediction of consumer behavior is crucial for optimizing marketing strategies. This study conducts a predictive analysis of consumer purchase intention using the UCL e-commerce platform dataset, employing systematic data preprocessing, feature engineering, and multiple machine learning models (including Support Vector Machine, K-Nearest Neighbors, and Decision Tree). Furthermore, we propose a hybrid approach that integrates deep learning techniques with the SVM model. Through comprehensive experiments, we demonstrate that high-dimensional data and nonlinear feature engineering significantly enhance prediction accuracy. Correlation analysis reveals “PageValues” and “ExitRates” as the most influential features affecting purchase decisions. Experimental results show that the proposed Deep Neural Network-enhanced SVM model achieves the highest prediction accuracy of 91.14%, outperforming baseline models. This research provides scientific guidance for e-commerce platforms to optimize marketing strategies and offers novel methodological insights for digital consumer behavior analysis.