摘要
为快速对数据进行特征选择以实现精确分类,采用M-distance算法思想进行数据集簇聚类,对样本数据进行预处理;设计加权K近邻算法缩减样本间距并构建样本分类模型;采用模拟简谐振动的方法遍历样本数据,求解最优加权特征向量,实现样本分类.实验结果表明:设计的算法是正确的,分类模型是合理的.在样本数据特征中,分离出的消费者最为关心的前10个样本特征符合消费者的行为选择,说明算法设计有一定实用性.
To select features quickly and classify the data accurately,the idea of M-distance algorithm is used to cluster the data set,and the sample data is preprocessed.The weighted K-nearest neighbor algorithm is designed to reduce the sample spacing and construct the sample classification model.The method of simulating harmonic vibration is used to traverse the sample data,solve the optimal weighted eigenvector and realize the sample classification.The experimental results show that the designed algorithm is correct and the classification model is reasonable.In the sample data features,the top 10 sample features that consumers are most concerned about are separated,which are in line with consumers′behavior choice,indicating that the algorithm design has a certain practicality.
作者
程勖
高雍政
郭芳
CHENG Xu;GAO Yongzheng;GUO Fang(School of Management, Dalian Polytechnic University, Dalian 116032, China)
出处
《大连理工大学学报》
CAS
CSCD
北大核心
2021年第6期645-651,共7页
Journal of Dalian University of Technology
基金
国家自然科学基金资助项目(71802035)
辽宁省纺织之光教学改革研究资助项目(BKJGLX333).
关键词
样本近邻
加权特征向量
谐振子
样本分类
sample nearest neighbor
weighted eigenvector
harmonic oscillator
sample classification
