<<  real data example (with normalisation)    index   

conclusions

• definitely can find interesting patterns in data
• challenge is to determine what the patterns represent; no guarantee a feature describes a class of document

todos

• run feature data through some classifiers, weka even, to see how well it does
• try some different normalisation schemes; tried tf/idf but it didn't seem much different to the unnormalised version

all the code is on github
my other random stuff can be found at matpalm.com

<<  real data example (with normalisation)    index