I have a matrix consisting of 10000 rows and 2 columns. I wanted to cluster my dataset into k number of clusters. I am aware of the K-mean available in MATLAB, but its clustering is based on first column.Any help will be a great help for me.
There is a great deal more even to MATLAB's pre-defined functions, although this may depend on what version/toolboxes you have available to you. Perhaps the following links may help:
Cluster Analysis: A Toolbox for MATLAB
(http://cda.psych.uiuc.edu/multivariate_fall_2013/5283-Millsap-Ch20.pdf)
Data Clustering [contains MATLAB code]
(http://people.whitman.edu/~hundledr/courses/M350S08/Chap09.pdf)
Some scripts that I think were based at least partly on An Introduction to Pattern Recognition: A MATLAB Approach:
http://mvhs.shodor.org/mvhsproj/clustering/MatClust2.html
However, I haven't examined them.
I am not sure I understand you.
This:
x=rand(10000,2);
k=5;
c=kmeans(x,5);
works in my Matlab (R2012b) without problem. As a result we get the numbers of cluster for which each data point is assigned.
@all,
My dataset included angle and distance. Angle was in degree and hence the value was not comparable with the distance (Angle was spanning 0-360 range while distance being in 0-4 range). I converted angle from degree to radian and Kmeans did the trick for me.
Thanx allo of you for your suggestions.
- Badges
- Science topic