You must use the oml.do_eval function to run a user-defined Python function.

The following example calls the errorModelPredict function, passing the test data without the target variable.

spatial_error_predictions = oml.do_eval(func='errorModelPredict', X=X_test.drop("MEDIAN_INCOME"))
print(spatial_error_predictions[:10])

The code prints the first ten predictions for the test data.

[[85565.81571662657], [88769.98209276547], [46010.46116330226], [61275.919165868865], [163674.5321011373], [40178.55663104116], [89290.25850064949], [47908.54834079923], [83884.02318889851], [50495.29040429841]]