Richard Edgar

Ahhh... OK. I was intending creating a separate `.rst` file, I can certainly merge the two together if the conclusion from the LRG discussion was to integrate things as much...

Should also have included this, which came from a discussion with some experts at Microsoft: 1. ~~Break it up into separate pages (as we are planning to do so anyway)~~...

I didn't want to tangle *too* much with "Fairness in Machine Learning" but I agree that it needs the same treatment overall.

Thank you for opening this issue! My immediate thought is that bootstrapping isn't really anything to do with `MetricFrame` - couldn't the approach you're describing be made into a wrapper...

@Harsha-Nori I agree that if a DataFrame-of-tuples is considered undesirable, then building this into `MetricFrame` is the proper approach. I do think that the statistics is going to need careful...

Having had a meeting with @MiroDudik and @Harsha-Nori , my problem was thinking about the issue backwards. I'm more used to the case of seeing people claim that two points...

Looking through old PRs, I noticed #908 ... Pickling is rather fragile; should we be adding save/load to `MetricFrame` ? That is, is this really a feature request, not a...

This is causing issues with some of our internal testing as well. We are investigating whether we can host the Census dataset on a CDN (there is retry logic there,...

Most likely, the retry logic was only in the AzureML Notebooks - there was never a significant problem here. Can you point me at a build where this has been...

So long as everything happens automatically, I'm not fussed.