Multivariate Anomalydetection

[mrks-g]

@mdbartos I was wondering how I need to construct the datapoint to use multivariate anomalydetection.

Currently I am using shingles of size 4 resulting in a datapoint of shape (1,4) e.g [[10,20,30,40]].
With a second feature would I do (2,4) or (1,8)?
Second was probosed by a colleague, but I do not get his point as this way, the second feature is not treated as a dimension.