Mining Association Rules for Optimizing Customer Purchase Behavior in E-Commerce Transactions

In a bid to improve e-commerce transactions and effective business strategies, it is important to understand customer purchase behavior. We introduced an integrated approach in this paper which was utilized to mine associate rules to optimize customer purchase behavior. Clustering algorithms (K-Means and DBSCAN) are used on the transactional data to clean the data by removing noise and dealing with missing values, which increases the quality of the data. The researchers analyze the frequent itemsets and association rules for co-occurrence, lift, and confidence measures using the Apriori, FP-Growth algorithms, AprioriDP, Eclat and RFrequent. Furthermore, classification models like Decision Tree and Random Forest are implemented for evaluation purposes and performance comparison. From the experimental results, the accuracy with decision tree was observed to be 72.54%, while random forest yields an accuracy of 77.48% with effective precision, recall and F1 score metrics. The paper also engages in comparative studies with recent studies to show the effectiveness of our method. These findings enable actionable insights for enhancing product suggestions and tailored marketing tactics in e-commerce.