prediction_intervals

mastml.plot_helper.prediction_intervals(model, X, rf_error_method, rf_error_percentile, Xtrain, Xtest)

Method to calculate prediction intervals when using Random Forest and Gaussian Process regression models.

Prediction intervals for random forest adapted from https://blog.datadive.net/prediction-intervals-for-random-forests/

Args:

model: (scikit-learn model/estimator object), a scikit-learn model object

X: (numpy array), array of X features

method: (str), type of error bar to formulate (e.g. “stdev” is standard deviation of predicted errors, “confint” is error bar as confidence interval

percentile: (int), percentile for which to form error bars

Returns:

err_up: (list), list of upper bounds of error bars for each data point

err_down: (list), list of lower bounds of error bars for each data point