An Intelligent Machine Learning Framework for Predicting Customer Purchase Decisions Using Classification and Regression Techniques in E-Commerce
Authors-S.Srinivas, Thumu Pradeepthi
Keyword-Customer Purchase Prediction, Machine Learning, Random Forest, Decision Tree, Logistic Regression, Feature Engineering, Classification, E-Commerce Analytics, Consumer Behavior, Predictive Modeling
The rapid expansion of e-commerce platforms has generated enormous volumes of consumer interaction data, creating new opportunities for understanding purchasing behavior through intelligent data analytics. Accurate prediction of customer purchase decisions enables online retailers to improve personalized recommendations, optimize promotional campaigns, and en-hance customer satisfaction. This study proposes a machine learning framework for predicting whether a customer will purchase a product after browsing it on an e-commerce platform. The research utilizes a large-scale transaction dataset obtained from JD.com containing customer information, product characteristics, browsing history, pricing details, and promotional activities. Before model construction, comprehensive data preprocessing and feature engineering tech-niques are performed to improve data quality and create informative predictive variables. The proposed framework evaluates three supervised machine learning algorithms, namely Decision Tree, Random Forest, and Logistic Regression, for binary purchase classification. Hyperparame-ter optimization is conducted using the Random Search strategy to maximize prediction perfor-mance while reducing overfitting. Model effectiveness is evaluated using Accuracy, Precision, Recall, and Area Under the ROC Curve (AUC). Experimental results demonstrate that the opti-mized Random Forest classifier achieves the highest prediction performance with a testing accu-racy of 0.999871 and an AUC of 0.9998, outperforming the remaining models. Feature im-portance analysis further indicates that coupon discount level and quantity discount level con-tribute most significantly to customer purchasing decisions. The proposed framework offers a reliable and computationally efficient solution for customer behavior prediction and provides valuable support for personalized recommendation systems, pricing optimization, inventory planning, and intelligent marketing strategies in modern e-commerce platforms.
An Intelligent Machine Learning Framework for Predicting Customer Purchase Decisions Using Classification and Regression Techniques in E-Commerce
Authors-S.Srinivas, Thumu Pradeepthi
Keyword-Customer Purchase Prediction, Machine Learning, Random Forest, Decision Tree, Logistic Regression, Feature Engineering, Classification, E-Commerce Analytics, Consumer Behavior, Predictive Modeling
The rapid expansion of e-commerce platforms has generated enormous volumes of consumer interaction data, creating new opportunities for understanding purchasing behavior through intelligent data analytics. Accurate prediction of customer purchase decisions enables online retailers to improve personalized recommendations, optimize promotional campaigns, and en-hance customer satisfaction. This study proposes a machine learning framework for predicting whether a customer will purchase a product after browsing it on an e-commerce platform. The research utilizes a large-scale transaction dataset obtained from JD.com containing customer information, product characteristics, browsing history, pricing details, and promotional activities. Before model construction, comprehensive data preprocessing and feature engineering tech-niques are performed to improve data quality and create informative predictive variables. The proposed framework evaluates three supervised machine learning algorithms, namely Decision Tree, Random Forest, and Logistic Regression, for binary purchase classification. Hyperparame-ter optimization is conducted using the Random Search strategy to maximize prediction perfor-mance while reducing overfitting. Model effectiveness is evaluated using Accuracy, Precision, Recall, and Area Under the ROC Curve (AUC). Experimental results demonstrate that the opti-mized Random Forest classifier achieves the highest prediction performance with a testing accu-racy of 0.999871 and an AUC of 0.9998, outperforming the remaining models. Feature im-portance analysis further indicates that coupon discount level and quantity discount level con-tribute most significantly to customer purchasing decisions. The proposed framework offers a reliable and computationally efficient solution for customer behavior prediction and provides valuable support for personalized recommendation systems, pricing optimization, inventory planning, and intelligent marketing strategies in modern e-commerce platforms.
Doi-[https://doi.org/10.5281/zenodo.21735797]