[Taken from https://en.wikipedia.org/wiki/Wikipedia talk:Web statistics tool; filed on behalf of EllenCT.]
Filtering transient spike anomalies
The WP:POPULARLOWQUALITY list suffers from the "known anomalies" problem of transient spikes.
There is a general solution to a closely related problem in the R code on pp. 191-2 of //Xu et al.// (2014) but I think it's much easier to use some measure of whether a spike less than four days long have standard deviations above, say +6, or one of the algorithms in http://stats.stackexchange.com/a/56744.
Can we get a separate API resource access for a filtered top-1000, please?