Is it intentional that some time series in NAB do not have regular time steps e.g. realAWSCloudwatch/ec2_cpu_utilization_825cc2.csv lines 39, 1116?
@dmac this would seem to be a problem for HTM because I guess it assumes regular sampling intervals or at least regular sampling patterns?
I’m not familiar withthat particular dataset.
One possible explanation is that there was a very long period of no-anomalies that the authors cut out of the dataset so that it would run faster.
That is just one example, 25 of the 58 time series in NAB have one or more irregular period.