Module panama.metrics.py_metrics

Functions

def nmae(y_true: numpy.ndarray | pandas.core.series.Series,
y_pred: numpy.ndarray | pandas.core.series.Series,
y_true_norm: numpy.ndarray | pandas.core.series.Series,
y_pred_norm: numpy.ndarray | pandas.core.series.Series,
normalization_mode: str) ‑> float

Expand source code

def nmae(
    y_true: Union[np.ndarray, pd.Series],
    y_pred: Union[np.ndarray, pd.Series],
    y_true_norm: Union[np.ndarray, pd.Series],
    y_pred_norm: Union[np.ndarray, pd.Series],
    normalization_mode: str,
) -> float:
    """
    Compute the normalized Mean Absolute Error (nMAE) between two arrays.

    Parameters:
        y_true (np.array, pd.Series): Array of true values.
        y_pred (np.array, pd.Series): Array of predicted values.
        y_true_norm (np.array, pd.Series): Array of normalization factors for y_true.
        y_pred_norm (np.array, pd.Series): Array of normalization factors for y_pred.
        normalization_mode (str): The normalization mode to use.
            'additive': Subtract the normalization factor from y_true.
            'multiplicative': Divide y_true by the normalization factor.

    Returns:
        float: normalized MAE value.
    """
    if normalization_mode.lower() == "additive":
        y_true = y_true - y_true_norm
        y_pred = y_pred - y_pred_norm
    elif normalization_mode.lower() == "multiplicative":
        y_true = y_true / y_true_norm
        y_pred = y_pred / y_pred_norm
    else:
        raise ValueError("normalization_mode must be 'additive' or 'multiplicative'")

    return mean_absolute_error(y_true, y_pred)  # type: ignore

Compute the normalized Mean Absolute Error (nMAE) between two arrays.

Parameters

y_true (np.array, pd.Series): Array of true values. y_pred (np.array, pd.Series): Array of predicted values. y_true_norm (np.array, pd.Series): Array of normalization factors for y_true. y_pred_norm (np.array, pd.Series): Array of normalization factors for y_pred. normalization_mode (str): The normalization mode to use. 'additive': Subtract the normalization factor from y_true. 'multiplicative': Divide y_true by the normalization factor.

Returns

float

normalized MAE value.

def nmape(y_true: numpy.ndarray | pandas.core.series.Series,
y_pred: numpy.ndarray | pandas.core.series.Series,
y_true_norm: numpy.ndarray | pandas.core.series.Series,
y_pred_norm: numpy.ndarray | pandas.core.series.Series,
normalization_mode: str) ‑> float

Expand source code

def nmape(
    y_true: Union[np.ndarray, pd.Series],
    y_pred: Union[np.ndarray, pd.Series],
    y_true_norm: Union[np.ndarray, pd.Series],
    y_pred_norm: Union[np.ndarray, pd.Series],
    normalization_mode: str,
) -> float:
    """
    Compute the normalized Mean Absolute Percentage Error (nMAPE) between two arrays.

    Parameters:
        y_true (np.array, pd.Series): Array of true values.
        y_pred (np.array, pd.Series): Array of predicted values.
        y_true_norm (np.array, pd.Series): Array of normalization factors for y_true.
        y_pred_norm (np.array, pd.Series): Array of normalization factors for y_pred.
        normalization_mode (str): The normalization mode to use.
            'additive': Subtract the normalization factor from y_true.
            'multiplicative': Divide y_true by the normalization factor.

    Returns:
        float: normalized MAPE value.
    """
    if normalization_mode.lower() == "additive":
        y_true = y_true - y_true_norm
        y_pred = y_pred - y_pred_norm
    elif normalization_mode.lower() == "multiplicative":
        y_true = y_true / y_true_norm
        y_pred = y_pred / y_pred_norm
    else:
        raise ValueError("normalization_mode must be 'additive' or 'multiplicative'")

    return mean_absolute_percentage_error(y_true, y_pred)  # type: ignore

Compute the normalized Mean Absolute Percentage Error (nMAPE) between two arrays.

Parameters

y_true (np.array, pd.Series): Array of true values. y_pred (np.array, pd.Series): Array of predicted values. y_true_norm (np.array, pd.Series): Array of normalization factors for y_true. y_pred_norm (np.array, pd.Series): Array of normalization factors for y_pred. normalization_mode (str): The normalization mode to use. 'additive': Subtract the normalization factor from y_true. 'multiplicative': Divide y_true by the normalization factor.

Returns

float

normalized MAPE value.

def rtae(y_true: numpy.ndarray, y_pred: numpy.ndarray) ‑> float

Expand source code

def rtae(y_true: np.ndarray, y_pred: np.ndarray) -> float:
    """
    Compute the Relative Total Absolte Error between two arrays.

    Parameters:
        y_true (np.array, pd.Series): Array of true values.
        y_pred (np.array, pd.Series): Array of predicted values.

    Returns:
        float: RTAE value.
    """

    return abs(y_true - y_pred).mean() / abs(y_true).mean()

Compute the Relative Total Absolte Error between two arrays.

Parameters

y_true (np.array, pd.Series): Array of true values. y_pred (np.array, pd.Series): Array of predicted values.

Returns

float

RTAE value.

def smape(y_true: numpy.ndarray | pandas.core.series.Series,
y_pred: numpy.ndarray | pandas.core.series.Series) ‑> numpy.float64

Expand source code

def smape(y_true: Union[np.ndarray, pd.Series], y_pred: Union[np.ndarray, pd.Series]) -> np.float64:
    """
    Computes the Symmetric Mean Absolute Percentage Error (SMAPE) between two arrays.

    Parameters:
        y_true (np.array, pd.Series): Array of true values.
        y_pred (np.array, pd.Series): Array of predicted values.

    Returns:
        float: Symmetric MAPE value.
    """
    return 1 / len(y_true) * np.sum(2 * np.abs(y_pred - y_true) / (np.abs(y_true) + np.abs(y_pred)))

Computes the Symmetric Mean Absolute Percentage Error (SMAPE) between two arrays.

Parameters

y_true (np.array, pd.Series): Array of true values. y_pred (np.array, pd.Series): Array of predicted values.

Returns

float

Symmetric MAPE value.