feat(bootstrap): generate confidence interval with bootstrap

[shenxiangzhuang]

Close #20

[shenxiangzhuang]

Please give some suggestions about this implementation and I'd like to improve it further.

[dustalov]

As mentioned in #20, I recommend taking a step back from rushing into implementing the idea and instead focusing on developing the API usage examples first.

It might be best to structure this as a wrapper function around the score computation function, ensuring it remains flexible and not tied to specific arguments passed to the estimator.

[shenxiangzhuang]

Hi @dustalov , I have reimplemented this feature according to the discussion in #20 (sure it need more friendly doc and tests, and I'll do these later:). Is there any suggestion for this until now?

ps: I found that using scipy.stats.bootstrap is not that straightforward as I thought before, so I just use np.quantile for for loop to do this now.

[dustalov]

Hi, apologies for the delayed review. While this implementation is straightforward, SciPy's version is intentionally more complex, likely for good reason. I would like to take some time to think it through—code in Evalica should provide clear added value. How urgent is this pull request?

[shenxiangzhuang]

Never mind, no rush~