Traffic and User Behavior Analysis of Online Mobile Games

Understanding user behavior of online mobile game user is quite important for online mobile game operators to improve user experience and increase revenue. In this paper, we present an approach to extract information of online mobile games from traffic records on network side and analyze user behavior based on the extracted information. Different from previous work, we use distributed computing platform including the Hadoop Distributed File System (HDFS) and the parallel processing software framework MapReduce. What's more, we analyze user behavior from four aspects: (i) activeness, (ii) diurnal variation, (iii) account distribution and (iv) version distribution. The analysis result based on the real data shows us some interesting conclusion that cann't be found before.