I am exploring azure multivariate detector services to apply on my multivariate timeseries having weekly periodicity.
Problem Statement: I have data having 105 rows(105 weeks) for one category.I have such 20k categories multivariate timeseries data with same 105 timestamps. I would like to train these data points for each category considering latest 8 weeks as sliding window to calculate the anomaly.
While applying azure multivariate detector, I realized that I can only select slidingwindow between 28 and 2880. When I selected the SlidingWindow=3, it raised below error:
"Error code: InvalidSlidingWindow Error message: The 'slidingWindow' field must be an integer between 28 and 2880"
I have gone through their documentation (https://learn.microsoft.com/en-us/azure/cognitive-services/anomaly-detector/concepts/best-practices-multivariate) and they mentioned that "When your data is periodic, you could set the length of 1 - 3 cycles as the slidingWindow".
I did not any find the reference how to implement "length of 1-3 cycles" for parameter slidingWindow
Can anyone help me on this ?