monotonicity check in Binomial Isotonic

[leongache]

Hi, thanks for sharing, highly appreciate. Can you check if BinomialIsotonicRegression should be checking for y monotonicity? is it necessary?

So, I'm running this: BinomialIsotonicRegression(10, increasing=True, curve_algo=PiecewiseConstantIsotonicCurve).fit(x, y) for probability calibration on binomial classifier where I make sure that x is monotonically increasing but can't have y to be increasing too as it takes 0 or 1 values corresponding to x.

Thanks