Numpy: mean/average absolute deviation mad()

It's just not that often used. statsmodels has a MAD implementation, please use that one.

Why should I use a different library just for one function. I do not understand why you refuse to add it while it is very easy to implement. It takes just a couple of lines!

Why should I use a different library just for one function.

That argument can be used for lots and lots of functions. we don't just add anything to numpy, it has to be useful to a wide enough audience. Robust statistics don't meet that bar; statsmodels or scipy.stats are the places to add such stats-specific functionality.

It takes just a couple of lines!

You can then very easily just define it yourself when you need it, if you don't want to use statsmodels.