基于LightGBM算法的游戏付费预测

Game Payment Prediction Based on LightGBM Algorithm

针对游戏内玩家付费情况预测的问题,通过对数据集的清洗、降维以及数据分析,对游戏内玩家ID、在线时长以及7 d内的付费情况等特征数据的提取和分析,并预测玩家前45 d的付费情况,通过4种算法的对比分析得出LightGBM模型的预测结果更加准确、效率更高。

In view of the problem of predicting the payment situation of players in the game,through the cleaning,dimensionality reduction and data analysis of the data set,the extraction and analysis of the characteristic data such as the player ID,online time and payment situation within 7 days in the game,the payment situation of the players in the first 45 days is predicted.Through the comparative analysis of four algorithms,it is concluded that the prediction result of LightGBM model is more accurate and more efficient.