基于改进K-Means算法的手机处理器聚类分析

Clustering Analysis of Mobile Processor Based on Improved K-Means Algorithm

摘要: 提出了一种改进的K-means聚类算法，并对2015至2019年手机市场主流的51款处理器进行聚类分析。首先使用手肘法改进K-means算法中K值的选取，求出最佳K值；其次利用欧氏距离求得各样本到聚类中心的距离，并将各样本归类到离其最近的聚类中心所在的簇中；重新计算新簇的聚类中心，若与旧聚类中心相同，则停止运算，否则重新计算各样本到新聚类中心的距离，重新归类直至新聚类中心与旧聚类中心相同；最后得到4个簇，分别包含7、16、14、14个样本，并将其分为高端、中端、中低端、低端处理器。

Abstract: An improved K-means clustering algorithm is proposed, and 51 mainstream processors in the mobile phone market from 2015 to 2019 are analyzed. Firstly, the elbow method is used to improve the selection of K value in k-means algorithm, and the best K value is obtained. Secondly, the Euclidean distance is used to find the distance from each sample to the cluster center, and all samples are gradually classified into the nearest cluster. Then the new cluster centers are recalculated. If the new cluster centers are the same as the old ones, the operation is stopped. Otherwise, the distance from each sample to the new cluster centers is recalculated and reclassified until the new cluster centers are the same as the old ones. Finally, four clusters are obtained, including 7, 16, 14 and 14 samples, which are divided into high-end, middle-end, middle-low-end and low-end processors according to the original data.