Benchmarking machine learning models for predictive analytics in e-commerce strategy

Predictive analytics is crucial in the digital economy, revolutionizing e-commerce by utilizing data insights to forecast trends, personalize customer experiences, and optimize operations with high accuracy (Gupta & Bansal, 2020; Jakkula, 2023). This research examines the comparative performance of traditional statistical techniques versus modern machine learning models in predicting customer behavior within Thailand’s e-commerce sector. Utilizing a comprehensive dataset from Thailand’s leading e-commerce platform, encompassing 10,000 customer interactions, this study analyzes the effectiveness of logistic regression, support vector machines (SVM), k-nearest neighbors (KNN), and random forest in modeling purchasing patterns. The findings demonstrate that SVM and KNN substantially outperform logistic regression in accuracy, precision, recall, and area under the curve (AUC), with random forest also showing significant capabilities in managing complex, large-scale datasets. This research highlights the critical role of advanced machine learning technologies in refining strategic decision-making within e-commerce by offering more accurate customer segmentation and enhanced targeting strategies. Given the swift growth of e-commerce in markets like Thailand, these insights provide crucial strategic implications for both local and international market contexts, suggesting a pivotal shift towards integrating machine learning to capitalize on the expansive digital consumer data available.