[Feature Request] Add Standard Deviation to summarystats()?
Feature Request: Add STD to summarystats()...
https://github.com/JuliaStats/StatsBase.jl/blob/a0e6f1e807a84a09b5f74431bb0099f4aaed5ae0/src/scalarstats.jl#L555
I agree that it's surprising not to print the standard deviation here. Do you feel like making a pull request to add it?
@nalimilan is this a dormant issue or still open for contribution?
I was thinking how about a summarystat() just like the summary function in R. It'll return different values for string/factor types just like R does...
We need more opinions to decide what's best. In DataFrames (https://github.com/JuliaData/DataFrames.jl/pull/2459), we decided not to report standard deviations and quartiles by default so that the output fits in the screen width: one needs to do describe(df, :detailed) to get them. Here screen width isn't a problem and we already report quartiles, so maybe we could print the standard deviation too.
@bkamins @pdeffebach What do you think?
In general I almost always want to see std, so I would like to have this change. The only issue is that it would be breaking. I am not sure what decision would be best. Maybe we can consider it to be mildly breaking and go for it?
I agree I almost always want std. I would maybe call it mildly breaking? Its really only useful in interactive work.
That would only change the printing, so that's considered non-breaking I think?
printing would be changed for describe, but summarystat is an object that stores the values (the struct would need to be changed, so e.g. if someone were serializing it it would break). See https://github.com/JuliaStats/StatsBase.jl/blob/master/src/scalarstats.jl#L858
Do we consider that adding a new field to an object is breaking though? That sounds quite restrictive.
OK - let us add it.
Will it be ok if I go on and try making a PR for this addition?
Sure.
Created #858