1

I'm trying to use Myrrix on a dataset, following the tutorial (using the web interface to ingest data). When I try the sample file (audioscrobbler-data.subset.csv) everything works as expected, but when I upload my own file it doesn't work, although it follows the same structure. I initially thought the issue may be values of 0, but removing those didn't help, and there's no error in the log - when I ingest, nothing seems to happen except for memory consumption going up for a minute.

See my input here: http://pastebin.com/F2dDQmFX

Sean Owen
  • 66,182
  • 23
  • 141
  • 173
nickb
  • 882
  • 3
  • 8
  • 22
  • So... what is the problem? nothing sounds wrong yet here. – Sean Owen Oct 23 '13 at 15:08
  • I would expect some log lines with iterations etc, same stuff that I see when uploading the audioscrobbler file, but nothing is printed to the log, and the "ready?" flag remains "false" – nickb Oct 23 '13 at 15:10
  • 1
    It doesn't rebuild a model after every input or upload, just when a threshold of input is reached. Is that the issue? force it manually with /refresh. – Sean Owen Oct 23 '13 at 15:13
  • Thanks! (for your awesome support, and for Myrrix too!) – nickb Oct 23 '13 at 15:18

1 Answers1

2

Just copying the comment for the record:

It will wait for a certain amount of input (which you can configure) before starting a new model build. You can suggest a new model build with the /refresh endpoint. But it will also be approximately updating the model in memory as it arrives anyway, so you don't necessarily have to force a rebuild in order to learn from the new data.

Sean Owen
  • 66,182
  • 23
  • 141
  • 173