I am looking to submit one of my own dataset to the NAB, in order to get the anomaly score with the numenta detector (thus I just want to use the --detect argument, as my dataset is not labelled).
I created a new directory in the data directory and put my dataset into it. Then I typed “python run.py -d numenta --detect” on the command window, but it doesn’t work…
Does anyone know how to do ?
You might need to supply a dummy windows file. See the section “Run subset of NAB data files” at the bottom of the NAB README.
The other option is to use HTM Studio. That will give you a nice visual output.
Thank you, it works for the NAB by supplying a dummy windows file and a dummy label.
And can the NAB work on categorical data ? I read somewhere in this forum that HTM studio didn’t, but maybe the NAB does ?
No, only a single stream of timestamped scalar data.