A comparative study of efficient initialization methods for the k-means clustering algorithm
A comparative study of efficient initialization methods for the k-means clustering algorithm
复制标题
DOI：
10.1016/j.eswa.2012.07.021
复制
发表时间：
2013-01-01
影响因子：
8.5
通讯作者：
Vela, Patricio A.
中科院分区：
文献类型：
--
作者：
Celebi, M. Emre;Kingravi, Hassan A.;Vela, Patricio A.
K-means无疑是使用最广泛的分区聚类算法。不幸的是，由于其梯度下降的性质，该算法是高度敏感的聚类中心的初始位置。已经提出了许多初始化方法来解决这个问题。在本文中，我们首先介绍了这些方法的概述，重点是它们的计算效率。然后，我们比较了八种常用的线性时间复杂度初始化方法的大型和不同的数据集，使用各种性能标准。最后，我们使用非参数统计检验分析实验结果，并提供建议，从业者。我们证明，流行的初始化方法往往表现不佳，事实上，这些方法有强大的替代品。(C)2012爱思唯尔有限公司保留所有权利。
K-means is undoubtedly the most widely used partitional clustering algorithm. Unfortunately, due to its gradient descent nature, this algorithm is highly sensitive to the initial placement of the cluster centers. Numerous initialization methods have been proposed to address this problem. In this paper, we first present an overview of these methods with an emphasis on their computational efficiency. We then compare eight commonly used linear time complexity initialization methods on a large and diverse collection of data sets using various performance criteria. Finally, we analyze the experimental results using nonparametric statistical tests and provide recommendations for practitioners. We demonstrate that popular initialization methods often perform poorly and that there are in fact strong alternatives to these methods. (C) 2012 Elsevier Ltd. All rights reserved.