摘要
机器学习中的决策树算法在处理没有属性缺失的数据集各样本时比较容易,但数据集较大时,往往某些属性会出现缺失值,这时就不能再使用通用方法来分析。利用决策树算法,以长白山植被识别为实例,通过对缺失值数据进行特殊处理,引入权重后重新计算每个属性的信息增益,再构建最优决策树,从而实现缺失值情况下对长白山植被的识别,能对新样本进行有效预测。
The decision tree algorithm in machine learning is easier to deal with the data set of which there is no attribute missing in each sample.However,when the data set is large,some attributes will have missing values.At this time,the general method can not be used for analysis.This paper uses the decision tree algorithm and takes Changbai Mountain vegetation identification as an example.By special processing to missing value data,the information gained from each attribute is recalculated after the weight is introduced,and then the optimal decision tree is constructed so as to realize the identification of Changbai Mountain vegetation under the missing value and give effective prediction to new samples.
作者
张华
许骏
付浩海
ZHANG Hua(School of Computer Technology&Engineering,Changchun Institute of Technology,Changchun 130012,China)
出处
《长春工程学院学报（自然科学版）》
2018年第4期80-84,共5页
Journal of Changchun Institute of Technology：Natural Sciences Edition
基金
吉林省科技厅项目(20170101009JC)
吉林省重点实验室支撑项目(20180622006JC)
吉林省教育厅项目(JJKH20191260KJ)
关键词
机器学习
决策树
样本
属性
缺失值
信息增益
machine learning
decision tree
sample
attribute
missing value
information gain
