本文關鍵詞： 添加變量 變量選擇 換機預測 Xgboost　出處：《數(shù)學的實踐與認識》2017年16期 　論文類型：期刊論文

【摘要】：首先對用戶數(shù)據(jù)進行特征分析,變量選擇,然后又采集了大量與手機性能相關的數(shù)據(jù)來擴充數(shù)據(jù)集,最后利用現(xiàn)代數(shù)據(jù)挖掘手段對用戶的換機行為進行預測,討論并比較了各種方法對換機預測的準確性.通過對用戶數(shù)據(jù)集進行測試實驗,表明變量選擇與補充能夠有效地提高移動用戶換機的預測結果,并且Xgboost方法在各種分析工具中的表現(xiàn)更為優(yōu)越.
[Abstract]:Firstly, the feature analysis and variable selection of user data are carried out, and then a large number of data related to mobile phone performance are collected to expand the data set. Finally, modern data mining methods are used to predict the behavior of users. This paper discusses and compares the accuracy of various methods to predict the change of computers. The test results of user data sets show that the selection and supplement of variables can effectively improve the prediction results of mobile users. And the Xgboost method is superior to other analytical tools.
【作者單位】： 太原理工大學數(shù)學學院;美國亞利桑那大學數(shù)學系;
【基金】：國家自然科學基金,高維數(shù)據(jù)變量間非線性交互作用的研究(11571009)
【分類號】：O212.1
，

本文編號：1532151