Predicting Customer Purchase Decisions using Data Mining Technique

• Main Authors: Dasig, Daniel D., Calantoc, Denver Jhon R., Guarin, Rudolph Val F., Taduyo, Mary Ann B., Ferrer, Christine N., Claricia, Eleonora E., Agus, Gina E.
• Format: Conference Proceeding
• Language: English
• Subjects: Behavioral sciences; customer purchase intention; Customer relationship management; Data mining; e-commerce; Electronic commerce; Humanoid robots; information density; k-means clustering; Knowledge discovery; Predictive models
• ISSN: 2770-0682
• DOI: 10.1109/HNICEM57413.2022.10109491

The k-means clustering algorithm was used to determine the clusters and patterns in Lazada and Shopee datasets retrieved from its websites using a spider bot as a scraping tool. Python, a multi-paradigm programming language, and the Knowledge Discovery in Databases (KDD) method were utilized in this study. Applying the elbow method, there were n=3 clusters created with Silhouette scores of 0.721 and 0.718, respectively. Two models were developed using Logistic Regression to predict the clusters of Lazada products and Shopee products with a coefficient of determination of 90.76 and 96.56, respectively. Consequently, product price, product ratings, and discounts influenced the customer's purchase intention. This study proved that information density and user reviews have greatly influenced consumer purchase decisions. Researchers recommend that empirical and analytical studies on information density and customer behavior, customer analytics be conducted to improve customer relationship management and apply other data mining techniques to improve the global marketplace architecture.