Predictive Analytics on Product Sales at Heva Inc. Using K – Means Method

Abstract

Prediction is the process of estimating something that is most likely to happen in the future based on previous and current knowledge that is owned, with the goal of minimizing the error. Prediction allows people to recognize and then solve difficulties that are occurring or are expected to arise. This study began with preparation, literature review, data collection, and knowledge discovery in databases (KDD). One of the processes is data mining using the K – Means method, which is critical for obtaining the research's results and conclusions. This research also uses the RapidMiner application as a comparison of the results with the results obtained by python coding. By using 4 clusters, products were categorized into 4 labels, namely very good products, good products, bad products, and very bad products. The research resulted in 11 products in the bad product category, 12 products in the good product category, 10 products in the very good category, and 18 products in the very good product category. The very good product label was further clarified with visualization to show the best time to restock each recommended product.