Sample problem results in `unexpected keyword argument` on current sklearn

[emunsing]

Similar to #13 and likely for the same reason: When running the sample problem in README.md on python==3.9.17, sklearn==1.1.1, irf==0.2.5 I get TypeError: __init__() got an unexpected keyword argument 'min_impurity_split' rising from the call to initialize WeightedDecisionTreeClassifier.

I'm new to the codebase, but has there been discussion of abstracting away the calls to the predictor class to the extent possible to allow for better flexibility with sklearn version? Or specify sklearn version in requirements.txt?