Abstract:
Market basket analysis (MBA) or affinity analysis, in the context of e-commerce, is a
business intelligence technique for predicting consumer buying decisions by reviewing
purchase history and generating association rules. It is connected to the detection
of hidden patterns in huge product databases. An effective analysis can improve a
company’s profitability, quality of service, and customer satisfaction. Therefore, the
main aim of this study is to find an efficient way to rate and recommend products based
on customer feedback and purchasing habits, as well as to gain a better understanding
of current customers’ behavior in order to predict future behavior and provide the best
product recommendation for their next purchase. As a result of this study, a superior
product recommendation system was developed using both MBA and textual/emoji
feedback from customers, which eliminated the drawbacks of existing systems. The
findings of this study show that analyzing both purchasing history and feedback leads
to improved product recommendations for customers’ next purchases. The Apriori
algorithm model was developed to discover patterns for implementing association rules
for recommender systems to find frequent item sets and significant relationships. The
minimal confidence and minimum support values used in mining rules to determine the
best-related product categories and items are the most important metrics. By analyzing
and evaluating feedback data, the customer satisfaction level was determined using
the Fast text embedding model. The results of the study reveal that the product
recommended appears to be more realistic and applicable because of the use of the
Apriori algorithm and Fast text embedding models.
