Predicting Next Purchase Item on JXM Game by K-Means Clustering and ARIMAX Model

Abstract

This paper is instructions for how to predict 5 next purchase items in “Vo Lam Truyen Ky Mobile” (JXM) which is a VNG's hot game. It's so hard to predict the next purchase item with a massive number of transactions per day. So, we analysed data to establish the solution to this problem. The solution make a prediction based on the past purchasing and in-game data with K-Means Clustering and ARIMAX model reaching accuracy about 70 percent. The research result can help to recommend items fit with users, improve user experience and may increase revenue.