35a9d4d6148f5933db0e3dfda23d8f0acf0403ba,AutoSklearn/components/classification/random_forest.py,RandomForest,get_hyperparameter_search_space,#,59

Before Change

        min_samples_split = scope.int(hp_quniform("min_samples_split", 1, 20, 1))
        min_samples_leaf = scope.int(hp_quniform("min_samples_leaf", 1, 20, 1))
        bootstrap = hp_choice("bootstrap", [True, False])
        return {"name": "random_forest",
                "n_estimators": n_estimators, "criterion": criterion,
                "max_features": max_features, "min_samples_split":
                min_samples_split, "min_samples_leaf": min_samples_leaf,
                "bootstrap": bootstrap}

    @staticmethod
    def get_all_accepted_hyperparameter_names():
        return (["n_estimators", "criterion", "max_features",

After Change

        // Don"t know how to parametrize this...RF should rather be
        // regularized by the other parameters
        // max_depth = hp_uniform("max_depth", lower, upper)
        max_depth = UnParametrizedHyperparameter("max_depth", "__None__")
        min_samples_split = UniformIntegerHyperparameter("min_samples_split",
                                                         1, 20)
        min_samples_leaf = UniformIntegerHyperparameter("min_samples_leaf",
                                                        1, 20)
        bootstrap = CategoricalHyperparameter("bootstrap", ["True", "False"])
        cs = ConfigurationSpace()
        cs.add_hyperparameter(n_estimators)
        cs.add_hyperparameter(criterion)
        cs.add_hyperparameter(max_features)
        cs.add_hyperparameter(max_depth)
        cs.add_hyperparameter(min_samples_split)
        cs.add_hyperparameter(min_samples_leaf)
        cs.add_hyperparameter(bootstrap)
        return cs

    @staticmethod
    def get_all_accepted_hyperparameter_names():
        return (["n_estimators", "criterion", "max_features",

Instances