Sorry, that "many_gyms" example is a half-baked example I never got around to finishing.
Ideally, you'll have one set of model params for each different field you want to predict. You can get each different model params by swarming over the same data, but specifying a different "predictedField" in the swarm description. You will have to name each model params file differently after the swarms. Then you create a model instance for each set of model params, each one predicting a different field. For each row of data, pass it into each model instance and get back a result from each. Each result will contain a prediction for a different field. --------- Matt Taylor OS Community Flag-Bearer Numenta On Tue, Aug 18, 2015 at 3:34 AM, Jorsde Del <[email protected]> wrote: > Hi > > i've been "playing" with the HotGym example and I'm trying to run on > multiple fields with no success. > > I've tried to follow and use > http://lists.numenta.org/pipermail/nupic_lists.numenta.org/2014-June/008691.html > http://lists.numenta.org/pipermail/nupic_lists.numenta.org/2015-February/010292.html > > In > https://github.com/rhyolight/nupic/tree/many-hot-gyms-prediction/examples/opf/clients/hotgym/prediction/many_gyms > (which I can't run) you are trying to run a comparison of multiple gyms , > but what if all you want to do is run a test on a single gym with many > variables ? > For example , KW consumption , Temp , Number of costumers , etc.. > > What need to be changed in the swarm desc in order to do it ? > > Thanks In Advance, > Jorsde
