I found that I need to change the setting in the run.py as well i.e., change the setting to 10 at line """ prediction = result.inferences["multiStepBestPredictions"][10] """. Predictions obtained were different from those obtained with setting of 1.
Based on the plot shown in the hot gym tutorial, predictions are one to one to the actual value. How to get the values beyond last date or timestamp ? That is, I want to see prediction when no actual value exist. Or predictions provided in the prediction column correspond to the forecast number of periods ahead. If this is the case then interpretation does not match with the plot in the tutorial shown Please advise From: Chittavajhula, Phanindra (MRO) Sent: Tuesday, March 29, 2016 4:03 PM To: 'nupic@lists.numenta.org' <nupic@lists.numenta.org> Subject: 10 Step prediction Hi All, I am using hot gym tutorial as the reference and trying to predict 10 steps ahead i.e., 10 days after the last day in my dataset. To simplify further, I replaced the data in the file with my own data, ran swarm and then ran run.py to get the predictions. I have the updated the swarm description file with the setting of 10 (predictionSteps = 10). However prediction seems to pick the default value of 1 and provides no prediction after the last day in the file. Are there any other places I need to update the code ? Can someone please advise ? Thanks, Phani
