Xây dựng mô hình dự đoán điểm chạm của khách hàng dựa trên dữ liệu hành trình mua sắm trực tuyến

The advent of digital platforms has profoundly transformed consumer behavior, necessitating advanced analytical approaches to understand the complex customer journey. This study addresses the critical need for businesses to leverage data-driven insights by focusing on building a robust predictive model for customer journey touchpoints and purchasing decisions within e-commerce websites. By dissecting online shopping behavior and demographic information, this research aims to significantly enhance personalized marketing strategies. The findings contribute to optimizing customer experiences and fostering more effective business engagements in the digital landscape.

E-commerce businesses, marketing strategists, data scientists, and academic researchers in marketing, business analytics, and artificial intelligence focusing on customer behavior and personalization.

This comprehensive study develops a sophisticated predictive framework designed to forecast customer journey touchpoints and subsequent purchasing decisions on e-commerce websites, with the ultimate goal of optimizing personalized marketing strategies. Utilizing a dataset comprising online shopping journey data and customer demographic information from the tourism sector collected between May 2015 and October 2016, the research employs a multi-stage methodology. Initially, K-Means clustering is applied to segment customers and identify target groups. Subsequently, Collaborative Filtering with Low Rank Matrix Factorization is used in conjunction with a Neural Network to predict the frequency of customer touchpoints. To predict purchase behavior, several machine learning models, including Logistic Regression, Decision Tree, Random Forest, KNN, and XGBoost, are then trained. The results definitively demonstrate that the Random Forest model achieves superior performance, exhibiting the highest accuracy of 96%, alongside strong F1-score and ROC-AUC metrics. Theoretically, this research makes a significant contribution by integrating process mining and recommendation systems for journey prediction, offering a novel approach to understanding and influencing customer behavior. Practically, it provides businesses with a tangible, data-driven model to gain deeper insights into customer interactions. Future research is encouraged to expand the prediction scope to all customer segments and to incorporate additional contextual factors, such as access devices and interaction duration, further enhancing personalization efforts and the overall efficacy of the website experience.