<![CDATA[The development of digital technology has encouraged the use of machine learning algorithms to understand and predict consumer behavior, particularly in the e-commerce sector. This study aims to build a predictive model for user purchasing decisions based on their interactions with digital product pages. The model was developed using a variant of the Decision Tree algorithm, namely C4.5, utilizing Gain Ratio as the best attribute selection criterion. The dataset used is synthetic and consists of several entries representing three main features: user visit duration, interaction with product reviews, and access to discount information. Training results show that the model is capable of producing good classification in identifying users who are likely to purchase the product. The feature "viewing reviews" is the most dominant attribute based on the highest Gain Ratio value, which is also the main node in the decision tree structure. Evaluation of the model's performance shows an increase in accuracy, precision, and recall after the application of the C4.5 algorithm, with the best accuracy reaching 88%. This study recommends further exploration using actual data and the addition of other behavioral variables to further improve the model's accuracy and generalizability in a real e-commerce context.]]>
Copyrights © 2025
