We have a mechanism that we use for unit tests to make the output of the random number generator predictable. Call RandomUtils.useTestSeed() before you do anything else and it should cause all the random number generation to follow the same pattern every time you run from the start. Is that what you need?
On Mon, Nov 23, 2009 at 3:36 PM, jamborta <[email protected]> wrote: > > hi, > > Just wondering if someone could suggest me a way how to fix the train/test > set and do not generate it every time I run the evaluation. I need to > evaluate some algorithms, and it would be important to use the same > train/test sets. > > thanks a lot > Tamas > > -- > View this message in context: > http://old.nabble.com/evaluating-recommender-systems---dividing-sets-tp26480189p26480189.html > Sent from the Mahout User List mailing list archive at Nabble.com. > >
