PROJECT TITLE :

Churn Prediction in Online Games Using Players’ Login Records: A Frequency Analysis Approach

ABSTRACT:

The rise of free-to-play and alternative service-based business models in the net gaming market delivered to game publishers problems typically associated to markets like mobile telecommunications and credit cards, especially client churn. Predictive models have long been used to handle this issue in these markets, where companies have a considerable amount of demographic, economic, and behavioral knowledge about their customers, while online game publishers usually only have behavioral knowledge. Easy time series' feature representation schemes like RFM can offer cheap predictive models solely primarily based on online game players' login records, but perhaps while not absolutely exploring the predictive potential of those information. We tend to propose a frequency analysis approach for feature illustration from login records for churn prediction modeling. These entries (from real data) were converted into fixed-length information arrays using four different strategies, and then these were used as input for training probabilistic classifiers with the k-nearest neighbors machine learning algorithm. The classifiers were then evaluated and compared using predictive performance metrics. One amongst the methods, the time-frequency plane domain analysis, showed satisfactory results, having the ability to theoretically increase the retention campaigns profits in a lot of than twentyp.c over the RFM approach.