层次聚类与K-means

Posted 2020-09-21 ColinQin

1. Hierarchical clustering(层次聚类)

作用：Clustering organizes things that are close into groups

算法步骤：

a).Find closest two thing

b).Put them together

c).Find next closest

     算法结果：A tree showing how close things are to each other

        

http://gallery.r-enthusiasts.com/RGraphGallery.php?graph=79

 

1. K-means

作用：Final estimate of cluster centroids

算法步骤：

a)Fix a number of clusters

b) Get “centroids” of each cluster

c)Assign things to closest centroid

d)Reclaculate centroids

     算法结果：按设定的cluster值分成相应和groups.