independent verification

[ZiluM]

Hi Feng,

I have updated the independent verification on CFR.
Please check it and merge it if possible.

Zilu

[CommonClimate]

Hi @ZiluM , thank you for implementing the changes.
I suggest a couple of minor modifications:

• the labels of the figure are wrong. For the lower panels, it should be CE (Coefficient of Efficiency), not Correlation efficiency.
• independent_verify() (infinitive form) is not very idiomatic. I would prefer indpdt_verif() or indpdt_verification(). Can you change the functions names in both .py modules and update the notebook?
• lastly, though it wasn't done in Hakim et al 2016, would it make sense to add the numbers of proxies in each class (assimilated/non-assimilated)? Presumably that would add up over all iterations -- I'm hoping that could be done in load_proxylabels()? It would give the audience a sense of how many samples those distributions are based on, which is always important. The other option is to use seaborn.histplot(stat='count') since seaborn is already a package dependency.

[fzhu2e]

@CommonClimate I've addressed the first two bullets in 668748e
please see the updated notebook: https://fzhu2e.github.io/cfr/notebooks/lmr-real-pages2k-indpdt-verif.html

Along with the commit, Python 3.13 is now supported and is the recommended version.

[CommonClimate]

Thank you so much @fzhu2e . Sorry my email client buries these notifications, so I just caught wind of it. As it happens, I was trying to do the modifications, but I hit a weird pandas bug that I tried to troubleshoot with Zilu, but never could. Anyway, I told @tanaya-g to play with the feature, and I hope she doesn't run into the same bug.
Good news on Python 3.13 as well!