posted 16 years ago
First, get a bunch of interesting data and put it in a database. Proper data mining has to have lots and lots of data.
Second, write Java code to access it.
The hard part is finding interesting connections in the data. This is essentially "cluster analysis" where you want to find in what dimensions, by what metrics, the data clumps together.
Also look at "Bayesian analysis" which can classify data by probabilistic interactions.
This is not easy stuff, but could make for a very interesting Senior project