decision_tree_optimization.py

from sklearn.datasets import load_iris
from sklearn.neighbors import KNeighborsClassifier
from sklearn.model_selection import cross_val_score

from hyperactive import Hyperactive, DecisionTreeOptimizer

data = load_iris()
X, y = data.data, data.target


def model(opt):
    knr = KNeighborsClassifier(n_neighbors=opt["n_neighbors"])
    scores = cross_val_score(knr, X, y, cv=5)
    score = scores.mean()

    return score


search_space = {
    "n_neighbors": list(range(1, 100)),
}

hyper = Hyperactive()
hyper.add_search(model, search_space, n_iter=100)
hyper.run()

search_data = hyper.results(model)

optimizer = DecisionTreeOptimizer(
    tree_regressor="random_forest",
    xi=0.02,
    warm_start_smbo=search_data,
    rand_rest_p=0.05,
)

hyper = Hyperactive()
hyper.add_search(model, search_space, optimizer=optimizer, n_iter=100)
hyper.run()