sinecode / MeanShift

C++ implementation of the mean shift clustering algorithm with OpenMP

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

question about bandwith

omerbrandis opened this issue · comments

hello

These are my first steps with mean shift, so please be gentle.

I've tried using msc on a very small dataset with 6 points
1,1,1
1,3,1
1,5,1
1,10,1
1,11,1
1,12,1

Can anyone help me calculate a suitable "bandwidth" value that will yield 2 clusters , points 1-3 in the first cluster , points 4-6 in the second.

thanks,
Omer.

Hi @omerbrandis, a bandwith of 2 will work:

 ./meanshift data.csv 2
File: data.csv
Number of points: 6
Number of dimensions: 3
Number of clusters: 2
Elapsed time: 2.5234e-05 s

Hello,

I'm pretty sure I tried using "2" back then,
retried it now and got :
Number of points: 6
Number of dimensions: 3
Number of clusters: 1

?
Omer.

I'm sorry but I can't reproduce.
Did you check the input file? I have this

$ cat data.csv
1,1,1
1,3,1
1,5,1
1,10,1
1,11,1
1,12,1
$ ./meanshift data.csv 2
File: data.csv
Number of points: 6
Number of dimensions: 3
Number of clusters: 2
Elapsed time: 3.1542e-05 s