FA-XGBoost model and precision medical prediction
首发时间:2019-02-25
Abstract:To predict scientifically and effectively on the increasing scale and dimension data with lack of some feature values, this paper proposes an XGBoost coupled with factor analysis model (FA-XGBoost), where factor analysis (FA) is used to reduce dimension of feature variables and then an XGBoost model is trained by using the data after FA. To test the model\'s effect, this paper analyzes some medical data, which are provided by the Tianchi Precision Medical Contest. The mean-square error (MSE) and the running time (t) are respectively 1.3800 and 1.3771 seconds for FA-XGBoost. Finally, we compared the FA-XGBoost model with four models based on decision trees. In general, GBDT and FA-XGBoost performed best on MSE, while FA-XGBoost worked best on running time.
keywords: Statistics XGBoost factor analysis blood glucose prediction diabete
点击查看论文中文信息
FA-XGBoost模型与精准医疗预测
摘要:为了在数据量和维度日益增长且数据存在缺失值的情况下科学有效地进行预测,本文提出了XGBoost耦合因子分析模型(FA-XGBoost),其中因子分析(FA)用于减少特征变量的维度,然后用因子分析降维后的数据训练XGBoost模型。为了验证模型的效果,本文分析了天池精准医疗大赛提供的一些医学数据。FA-XGBoost的均方误差(MSE)和运行时间(t)分别为1.3800和1.3771秒。最后,我们将FA-XGBoost模型与基于决策树的四个模型进行了比较。总体上,GBDT和FA-XGBoost在MSE上表现最佳,而FA-XGBoost在运行时表现最佳。
基金:
引用
No.****
动态公开评议
共计0人参与
勘误表
FA-XGBoost模型与精准医疗预测
评论
全部评论0/1000