For a given set of samples, a spike (a single sample that is inconsistent with the signal in general), can indicate a faulty measurement or other anomaly. It is very useful to analyze data with the spikes removed.
Thanks you for opening this issue @inselbuch! To check: are you talking about techniques such as those referenced in Analytical functions for Anomaly Detection issue, or did you have something different in mind?
For a given set of samples, a spike (a single sample that is inconsistent with the signal in general), can indicate a faulty measurement or other anomaly. It is very useful to analyze data with the spikes removed.