out_of_bounds

  • Available in: Isotonic Regression

  • Hyperparameter: no

Description

Use this option to specify how a trained model should treat values of an X predictor that are outside of the bounds seen in training.

Available options for out_of_bounds include the following:

  • na: Output NA for values that are outside of the interval seen during training. This is the default option.

  • clip: Use the prediction of the smallest or largest seen value depending on what side of the training interval the particular value falls in.

Example

import h2o
from h2o import H2OFrame
from h2o.estimators.isotonicregression import H2OIsotonicRegressionEstimator
import numpy as np
from sklearn.datasets import make_regression
h2o.init()

X_full, y_full = make_regression(n_samples=10000, n_features=1, random_state=41, noise=0.8)
X_full = X_full.reshape(-1)

p05 = np.quantile(X_full, 0.05)
p95 = np.quantile(X_full, 0.95)

X = X_full[np.logical_and(p05 < X_full, X_full < p95)]
y = y_full[np.logical_and(p05 < X_full, X_full < p95)]

train = H2OFrame(np.column_stack((y, X)), column_names=["y", "X"])
h2o_iso_reg = H2OIsotonicRegressionEstimator(out_of_bounds="clip")
h2o_iso_reg.train(training_frame=train, x="X", y="y")

test = H2OFrame(np.column_stack((y_full, X_full)), column_names=["y", "X"])
h2o_test_preds = h2o_iso_reg.predict(test).as_data_frame()
print(h2o_test_preds)