VineRegressor.predict

VineRegressor.predict(X)

Predicts the conditional mean and/or quantiles of Y given X.

Computes weights \(w_i(x)\) from the fitted copula (_iter_weights) and returns the weighted statistics: \(\hat{\mathbb{E}}[Y \mid X = x] = \sum_i w_i(x)\, y_i\) for the mean (closed-form solution of the estimating equation \(\int (y - \beta) \hat f(y \mid x)\, dy = 0\)) and the weighted quantile via numpy.quantile() with method="inverted_cdf" for each requested level.

Parameters:

Xndarray, shape (n_samples, n_features), dtype float, or DataFrame

Test covariates. Must match the training schema.

Returns:

ndarray, shape (n_samples,) or (n_samples, n_outputs), dtype float

Predictions. Shape (n_samples,) if only one output is requested (mean or a single quantile), otherwise (n_samples, n_outputs). Output columns are ordered: mean (if self.mean), then quantiles in self.quantiles order.