# classification machinery needed

Consider a set of 7D vectors. Each vector belongs to one of four classes.

After mapping to 3D with PCA and coloring each point according its class the dataset looks like as shown below:

For the animated GIF version, click here https://docs.google.com/open?id=0B26-1RYIJR_oOWRmSXZyWUl4M2M

What classification technique would be suitable for this dataset (e.g. decision trees, neural networks, etc) ?

-
What space are you working in? $R^7$? Or a discrete space? Am a bit puzzled by your remark: "coloring each point according to its class". This suggests that you already have a classifier? –  Assad Ebrahim Dec 26 '12 at 23:15

## 1 Answer

While PCA is useful to reduce dimensions, the choice of the number of dimensions should be viewed as a parameter depending on the effectiveness of your classifier on the data generated by your problem. (It could be three, two, four, ...)

But then you'll need a classifier, and you might get good mileage from one of the various mathematical segmentation / clustering algorithms. Two that are popular and fairly easy to implement are K-Means Clustering and Fuzzy C-Means Clustering. There are lots of existing implementations for these in Matlab, Octave, or R.

The following links should get you started:

-