I think I got it now that the angle in this case is somewhat superfluous and in fact we're giving nupic a stream of sine values and asking nupic to predict the next sine value.
On Sun, Nov 30, 2014 at 9:15 PM, J <[email protected]> wrote: > > New to nupic here. Going through the "Predicting Sine Waves with NuPIC" > https://www.youtube.com/watch?v=KuFfm3ncEwI video. > > For the following code: > > 61 with open(file_path, "rb") as input_file: > 62 reader = csv.reader(input_file) > 63 > 64 #skip header rows > 65 reader.next() > 66 reader.next() > 67 reader.next() > 68 > 69 for row in reader: > 70 angle = float(row[0]) > 71 sine_value = float(row[1]) > 72 result = model.run({'sine': sine_value}) > 73 output.write(angle, sine_value, result, prediction_step=1) > > This may just me being dense but my understanding was that the angle was > "x" and we were trying to predict the "y" which is sin(x). So why do we > pass the y into nupic rather than the x? I would have expected that we > would give nupic an x independent variable (the angle in this case) and get > back a y dependent variable (the result of sin(x)). Why are we not passing > the angle to the model.run method? > > result = model.run({'sine': angle}) > > What am I missing? > >
