Interquartile range about bootstrap.jl

[juliangehring]

Okay, happy to hear that it is working fine for you.
You can use an arbitrary estimator function with the package, as in this case coming from the StatsBase package. The bootstrap package doesn't distribute any particular statistics function and is designed to work with whatever estimator function that you choose, irrespectively of where the function is defined.

[AhmedAlreweny]

It is very clear now. Thanks a lot.

[juliangehring]

Could you share some more details about the idea? Where do you see the benefits?

[AhmedAlreweny]

It is an estimator that measures how the data are deviated from the median (similar to the std). The difference is that the std can be misleading if the distribution has outliers or heavy tail. In that case, the mean and std are replaced with the median and the iqr, respectively.

[juliangehring]

I was rather asking about where you think that the iqr is "missing" in the package and in which form it should be "added".

You can already use the iqr as an estimator:

bootstrap(StatsBase.iqr, data, BootstrapSampling)

Doesn't that work as expected?

[AhmedAlreweny]

Oh! I am sorry. I tried bootstrap(iqr, data, BootstrapSampling) and when it didn't work, I thought that it is not included. It is working fine. Thanks a lot!