Open Antsypc opened 2 years ago
No it's not a mistake. Those two lines change the initial values in each stream (of length config.l_s
) to the mean of the initial values in the stream to allow a small amount of historical context to be collected prior to attempting to identify anomalies. This is done for all channels except for C-2
, which has an anomaly early in its time series. In this instance we don't set early values to the mean so this early anomaly has a chance to be detected.
It seems like
if not channel.id == 'C-2'
, code here, should beif channel.id == 'C-2'
. Is this a mistake? Thanks.