This is a NuPIC bug, anomaly score should be float, same for multistep.predictions.N, before they settle, for the 1-N rows the value is None. This complicates the processing a bit here, we could use 0.5 for anomaly and say 0.0 for predictions during the burn in period.
https://github.com/nupic-community/nupic.visualizations/blob/master/client/src/app/appCtrl.js#L110-L113
This is a NuPIC bug,
anomaly score
should be float, same formultistep.predictions.N
, before they settle, for the 1-N rows the value isNone
. This complicates the processing a bit here, we could use 0.5 for anomaly and say 0.0 for predictions during the burn in period.This should be resolved upstream.