Huge Data Set Alghoritm Question

[tibun]

Hello,
I am trying to bite huge data set mining problem, I have 5 different types of classifiers that are being identified by values in 5 columns.

The problem is that the column values are pretty much close to each other, there is lots of values that are similar or different by a fraction and they were classified differently.

Hot to bite such a problem? I tried Random Trees, Clustering Algorithms, - maybe an advice which one is the best in this case, or perhaps I should use more tight classifications that would identify the chain of values from this 5 columns in a more efficient way?

Hope it is clear enough - thanks in advance.