I’ll start by saying that I understand why I’m getting the result that I am, with some readings picking up or losing one data point compared to most of the others.
I’ve found two methods to get rid of the spikes, but I dont feel that either are good solutions. One is to set the group by time interval to 1 second and fill to null, but this doesn’t work at all for timescales beyond a few hours. The other is to have a very wide moving average, which would hide the spikes as well as real anomalies that I want to see.
Are there any other solutions? I feel like using the mean aggregation should take care of it, that way having two or three datapoints in the same window would get averaged out. but it doesn’t seem to do anything at all.