基于机器学习统计思想实现多重线性回归分析  

Realization of a multiple linear regression analysis based on the machine learning statistical thought

作　　者：谷恒明 胡良平[1,2] 

机构地区：[1]军事医学科学院生物医学统计学咨询中心,北京100850 [2]世界中医药学会联合会临床科研统计学专业委员会,北京100029

出　　处：《四川精神卫生》2018年第1期15-18,共4页Sichuan Mental Health

基　　金：国家高技术研究发展计划课题资助(2015AA020102)

摘　　要：本文目的是介绍基于机器学习统计思想实现多重线性回归分析的方法。首先,简要回答了几个基本问题:什么是机器学习、它能解决的统计学问题及其具体方法;进一步,还粗略地介绍了BP神经网络回归分析方法的基本思路;最后,通过一个实例详细呈现了如何基于R软件实现BP神经网络回归分析的全过程。The purpose of this paper was to introduce a method based on the machine learning statistics to implement a multiple linear regression analysis. First of all,it answered several basic questions briefly,such as what machine learning is,what statistical problems it can solve,what are its specific methods. Further,the brief introduction of the basic idea of BP neural networks regression analysis was presented. Finally,an example was given to show how to implement the whole process of BP neural networks regression analysis based on R software.

关 键 词：回归分析 机器学习 误差反向传播神经网络 R软件 输入节点 隐节点 输出节点 

分 类 号：R195.1[医药卫生—卫生统计学]