In case of 2-class data, it would be nice to be able to choose a different cutoff point. I.e., in model.predict() compare the decision_function values against a value different than 0.
In order to make the chosen values generalizable, the values returned by decision_function might need to be standardized to a fixed interval (they are currently unbound).
In case of 2-class data, it would be nice to be able to choose a different cutoff point. I.e., in model.predict() compare the decision_function values against a value different than 0. In order to make the chosen values generalizable, the values returned by decision_function might need to be standardized to a fixed interval (they are currently unbound).