Akademik Veri Yönetim Sistemi
6th International Conference on Statistics: Theory and Applications, ICSTA 2024, Barcelona, İspanya, 19 - 21 Ağustos 2024
In the recent years, an increase in significance of e-commerce resulted from a wide use of cellular apparatuses and development of social media. In effort to move with time, many consumers shop using their cellular gadgets; so, these platforms are important for organizations when seeking to market themselves and improve on profits. But there are other challenges in e-commerce including security concerns, increased competition, logistical problems, customer satisfaction levels and issues related to brand names. This paper is about predicting whether customers will leave online shopping services through machine learning approaches especially using Logistic Regression and Decision Tree algorithms. Various measures such as confusion matrix, F1 score, cross-validation accuracy, precision, recall and ROC curve were employed in evaluating the models. An excellent 86% cross-validation accuracy was observed with Decision Tree algorithm, suggesting its better performance relative to other algorithms that were tested.