5
$\begingroup$

I'm testing the latest Mathematica available on the Raspberry Pi (10.3.1 for Linux ARM (32-bit) (January 11, 2016)) for machine learning but I'm having issues with the default behaviour of Predict.

Even the simplest examples like this one will hang, or at least take more than one hour:

Predict[{1->1, 2->2, 3->3}]

Selecting another PerformanceGoal for Predict (like "TrainingSpeed" or simply "Speed") will fix the problem, but the default option seems to have some issues. Classify seems to work fine with default options.

Is this a bug of Predict or simply for the RPi? Setting PerformanceGoal to "TrainingSpeed" will significantly degrade the prediction compared to "Quality" or Automatic?

$\endgroup$
2
  • $\begingroup$ I have experienced the same issue $\endgroup$
    – M.R.
    Commented Jan 1, 2017 at 1:23
  • 1
    $\begingroup$ This should be improved in the next release -- the given Predict example takes about 11 seconds for me on a Raspberry Pi 3. $\endgroup$
    – ilian
    Commented Jan 1, 2017 at 17:13

1 Answer 1

2
$\begingroup$

This should be improved in the just released Mathematica 11.0.1 for the Raspberry Pi, for example

$Version

(* "11.0.1 for Linux ARM (32-bit) (January 17, 2017)" *)

AbsoluteTiming[Predict[{1 -> 1, 2 -> 2, 3 -> 3}]]

(* {12.3929, PredictorFunction[...]} *)
$\endgroup$
2
  • $\begingroup$ 67 seconds on a Raspberry Pi 1 (for comparison). RPi 3 seems much faster. $\endgroup$
    – Szabolcs
    Commented Feb 24, 2017 at 21:21
  • $\begingroup$ Most time seems to have been taken by loading. Second run took only 22 seconds. $\endgroup$
    – Szabolcs
    Commented Feb 24, 2017 at 21:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.