# Mean Absolute Error (MAE) calculation
[docs]
def mean_absolute_error(y_true, y_pred):
"""
Calculate the Mean Absolute Error (MAE) metric for regression.
This function computes the average absolute difference between the predicted values (`y_pred`)
and the actual values (`y_true`). It measures the magnitude of errors in prediction, providing
a straightforward evaluation of a model's accuracy.
Parameters:
----------
y_true : array-like
True values of the target variable.
y_pred : array-like
Predicted values from the model.
Returns:
-------
float
The Mean Absolute Error.
Notes:
------
MAE is defined as:
MAE = (1 / n) * sum(|y_true - y_pred|)
where n is the number of observations.
Examples:
---------
>>> y_true = [100, 200, 300]
>>> y_pred = [110, 190, 290]
>>> mean_absolute_error(y_true, y_pred)
10.0
"""
# Ensure the input lists have the same length; otherwise, raise an error
if len(y_true) != len(y_pred):
raise ValueError("The lengths of y_true and y_pred must be the same.")
# Compute the absolute differences between actual and predicted values, sum them,
# and divide by the total number of observations to calculate MAE
mae = sum(abs(true - pred) for true, pred in zip(y_true, y_pred)) / len(y_true)
return mae