I usually only use the inferenceShifter when I am graphing predictions and real data on the same chart. Shifting in this fashion ensures the data and predictions are properly aligned at each time marker.
--------- Matt Taylor OS Community Flag-Bearer Numenta On Wed, Sep 2, 2015 at 8:16 AM, Austin Marshall <[email protected]> wrote: > InferenceShifter is used to align predictions with actual values for > comparison, hence its usage inside the if plot: block. It's especially > useful for multistep prediction models where there might be, for example, > predictions for 1, 5, and 10 steps ahead. > > On Wed, Sep 2, 2015 at 7:15 AM, 曾惟如 <[email protected]> wrote: >> >> hello nupic: >> >> recently, I am learning the source code of the program hotgym prodiction. >> and there are something I can't understand. >> >> In the document of Online Prediction >> Framework(https://github.com/numenta/nupic/wiki/Online-Prediction-Framework), >> there a introduction(Shifting Inferences) which is about the shift, but i >> can't get what dose that mean. So I want to ask some question. >> >> 1: what dose shifter do?. What dose the code "result - >> shifter.shift(result)" do?. You can see some codes below. >> >> 2: Why some program do not use the class InferenceShifter? I have read >> some other source code like One Hot Gym Anomaly and CPU Usage. in that code, >> there no the step "shifter = InferenceShifter()", it dosen't use the class >> InferenceShifter. So i want to konw what is situation of use the class >> InferenceShifter. >> >> "field=kw_energy_consumption"]) >> >> if plot: >> result = shifter.shift(result) >> >> prediction = result.inferences["multiStepBestPredictions"][1] >> output.write([timestamp], [consumption], [prediction]) >> >> (the code is in the line 124 of the hotgym_prediction's run.py) >> >> Please bear the bad English, andthanks in advance!! > >
