Method sample_predictive_trend_vectorized

python/prophet/forecaster.py:1634–1664

Sample draws of the future trend values. Vectorized version of sample_predictive_trend(). Parameters ---------- df: Prediction dataframe. iteration: Int sampling iteration to use parameters from. n_samples: Number of future paths of the trend to simulate.

(self, df: pd.DataFrame, n_samples: int, iteration: int = 0)

        return trend * self.y_scale + df['floor']

    def sample_predictive_trend_vectorized(self, df: pd.DataFrame, n_samples: int, iteration: int = 0) -> np.ndarray:
        """Sample draws of the future trend values. Vectorized version of sample_predictive_trend().

        Parameters
        ----------
        df: Prediction dataframe.
        iteration: Int sampling iteration to use parameters from.
        n_samples: Number of future paths of the trend to simulate.

        Returns
        -------
        Draws of the trend values with shape (n_samples, len(df)). Values are on the scale of the original data.
        """
        deltas = self.params["delta"][iteration]
        m = self.params["m"][iteration]
        k = self.params["k"][iteration]
        if self.growth == "linear":
            expected = self.piecewise_linear(df["t"].values, deltas, k, m, self.changepoints_t)
        elif self.growth == "logistic":
            expected = self.piecewise_logistic(
                df["t"].values, df["cap_scaled"].values, deltas, k, m, self.changepoints_t
            )
        elif self.growth == "flat":
            expected = self.flat_trend(df["t"].values, m)
        else:
            raise NotImplementedError
        uncertainty = self._sample_uncertainty(df, n_samples, iteration)
        return (
            (np.tile(expected, (n_samples, 1)) + uncertainty) * self.y_scale +
            np.tile(df["floor"].values, (n_samples, 1))
        )

    def _sample_uncertainty(self, df: pd.DataFrame, n_samples: int, iteration: int = 0) -> np.ndarray:
        """Sample draws of future trend changes, vectorizing as much as possible.