Design And Implementation Of A Game Precision Marketing System Based On Data Mining

While the Internet game industry growing and its demographic dividendgradually disappearing, the game industry is turning into a commercial Red Sea.Meanwhile, precision marketing has become an important factor of success and costreduction in game industry. More companies are searching insights from the datathey posess. One of the core technologies in precision marketing is target usersidentification in different senarios. The traditional statistical methods havelimitations gaining useful insights out of data. In contrast, data mining is able toanalyze the big data, gain deep insights and complex patterns. With improvement ofcomputing and mathematics, data mining is becoming an important method toexplore user behavior patterns and find out target users. However, there are stillmany difficulties in applying data mining into the actual game marketing, such asthe absence of suffient practice solutions.This paper is a study on how to better introduce data mining to game marketing.First, it analyzes the forms of game marketing activities and the needs of the freegame marketing in Tencent. The core technical issue is then summarized into a userclassification problem. Second, based on the practice and the CRISP-DM(cross-industry standard process for data mining) model, this paper defines thescope of both manual responsibilities and the programming tasks. Then this paperdesigns and implements a game precision marketing system based on data mining,including the following main components.1. The feature analyzing tool-correlationanalysis between player behavior attributes and marketing objectives.2. The featureselection tool-decision tree algorithm to generate attribut es weight outputs.Analysts can re-select the features on the basis of attributes weights.3. Themodeling and assessment tool-classification algorithm and the data storage systeminterface. Analysts could use it for testing and assessment.4. The automatic userclassification system. It implements a scheduling and result statistical system abovethe modeling and assessment tool.5. The advertisement inquiry system. It’s aunified management of target users predicted by the automatic user classificationsubsystem. It provides ads inquiry service to marketing department and helps to optimize the advertising process.The system had access to many Tencent games after launcing. In the case of"Cross Fire", this system brought300,000paid subscribers in9marketing activitiesin9months. The average increasing rate per phase was220%.The feature analyzing and feature selection tool helped impove the businessprocess. It also helped reduce the human resource cost from1person per month to1person per week. After launching the automatic user classification system, the gamemarketing activities of certain types will no longer need any labor input, whichproves that the game precision marketing system based on data mining really caneffectively improve profit and efficiency.