Isolation Forest: Combining Input Features with Output Y

Question

Seems to me that when you have a data-set with outliers, those outliers might be caused by not only the features, but by output variable as well.

So for anomaly detection in your training set, why does every example i see on the internet NOT include features appended with output variable?

Answer 1

actually you can approach an anomaly like that, but there are some problems outliers are varied in features that may cause a lot of problems here and most of the advanced anomaly models are the online trainer so it may cause a lot of afford to do this your model will be too much complicated I draw an example for you