Hi,
I also looked at TestOutliers function because it is just below the NewData function. I am not sure whether an absolute limit to call a value an outlier fits all the needs, especially when one population might be very abundant or very rare. I imagine the current threshold of 4 MAD will rarely raise an alert. You probably thought about adjusting the threshold according to the number of events. If so could tell for which reason you didn't retain this approach?
As this view sounds familiar to my mind, maybe we already discussed about it. Sorry if this is the case.
Best,
Samuel
Hi, I also looked at TestOutliers function because it is just below the NewData function. I am not sure whether an absolute limit to call a value an outlier fits all the needs, especially when one population might be very abundant or very rare. I imagine the current threshold of 4 MAD will rarely raise an alert. You probably thought about adjusting the threshold according to the number of events. If so could tell for which reason you didn't retain this approach? As this view sounds familiar to my mind, maybe we already discussed about it. Sorry if this is the case. Best, Samuel