Boosting类算法比较研究——以线上优惠券回收情况预测为例  

Comparative Research on Boosting Algorithms:Taking Online Coupon Recovery Prediction for Example

作　　者：翟大昆[1] 牟文龙 ZHAI Dakun;MOU Wenlong(School of Business and Tourism Management,Yunnan University,Kunming Yunnan 650091)

机构地区：[1]云南大学工商管理与旅游管理学院,云南昆明650091

出　　处：《软件》2021年第10期67-69,共3页Software

摘　　要：作为集成学习的重要组成部分,Boosting类算法目前在机器学习领域受到广泛的关注。为了对Boosting类算法的性能进行比较分析,本文以线上优惠券回收预测模型的研究为例,对比Adaboost、GBDT、XGBoost三种主流Boosting算法在该分类问题上的拟合效果。并在其中选取拟合效果最好的算法代表Boosting类算法,与传统机器学习模型的拟合效果进行对比分析。最终发现XGBoost在线上优惠券回收预测模型中的表现优于另外两种算法。As an important part of ensemble learning,Boosting algorithms are currently receiving extensive attention in the field of machine learning.In order to compare and analyze the performance of Boosting algorithms,this paper takes the study of online coupon recovery prediction model as an example to compare the fitting effects of the three mainstream Boosting algorithms Adaboost,GBDT,and XGBoost on this classification problem.The algorithm with the best combined effect represents the Boosting algorithm and compares with the traditional machine learning models.Ultimately,it is found that the performance of XGBoost is better than the other two algorithms on the stduy of online coupon recovery prediction model.

关 键 词：性能分析 线上优惠券回收预测模型 Boosting类算法 极限梯度提升算法 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]