How to write a custom wrapper for a prediction function in xgboost or other estimators

Question

So I want to manipulate the result of my prediction and I need to do it within the estimator. I tried to write a wrapper like this, but my kernel just dies when I execute the predict function. From my understanding this should just replace the predict function in xgboost right?

from xgboost import XGBRegressor as xgb


class custXGB(xgb):

   def predict(self, X, y=None):
        return self.predict(X)

I then fit the clas normally but when I use predict the kernel dies without error:

estimator = custXGB()
estimator.fit(X_train, y_train)
# works fine

estimator.predict(X_train)
#kernel dies

Answer 1

You've written an infinite loop: your predict is calling itself, not the XGBRegressor.predict. Replace self with super() inside the method.