cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
Check out the JMP® Marketplace featured Capability Explorer add-in
Choose Language Hide Translation Bar
likunz0
Level I

What distance is saved when I click "save clusters" in the K-means clustering report

I am wondering what distance is saved for each row when I click "save clusters" in the Kmeans clustering report. I used the K-means method to participate my data table of 10000 rows into 50 clusters. When I clicked "save clusters", I saved two columns. One is the cluster column, which indicate which cluster the row is assigned to; the other one is called "Distance". I am wondering what distance is the "Distance". I found that the distance between each row and the cluster center  is much smaller than the "Distance".

2 REPLIES 2
txnelson
Super User

Re: What distance is saved when I click "save clusters" in the K-means clustering report

This is taken from the Multivariate Methods book available in JMP under Help==>Books==>Multivariate Methods11050_Documentation on Distance.jpg

Jim
likunz0
Level I

Re: What distance is saved when I click "save clusters" in the K-means clustering report

Thanks, Jim. It looks like that the distance is calculated as the Euclidean length between two vectors. Then which two vectors are used to calculate the "Distance"? Is it the distance between each row and the center of the cluster of that row? Or is it the distance between each row and the mean of all rows? However, rrom my calculation, the "Distance" is larger than both cases.