|
| 1 | +""" |
| 2 | +Mean Squared Logarithmic Error (MSLE) Loss Function |
| 3 | +
|
| 4 | +Description: |
| 5 | +MSLE measures the mean squared logarithmic difference between |
| 6 | +true values and predicted values, particularly useful when |
| 7 | +dealing with regression problems involving skewed or large-value |
| 8 | +targets. It is often used when the relative differences between |
| 9 | +predicted and true values are more important than absolute |
| 10 | +differences. |
| 11 | +
|
| 12 | +Formula: |
| 13 | +MSLE = (1/n) * Σ(log(1 + y_true) - log(1 + y_pred))^2 |
| 14 | +
|
| 15 | +Source: |
| 16 | +(https://insideaiml.com/blog/MeanSquared-Logarithmic-Error-Loss-1035) |
| 17 | +""" |
| 18 | + |
| 19 | +import numpy as np |
| 20 | + |
| 21 | + |
| 22 | +def mean_squared_logarithmic_error(y_true: np.ndarray, y_pred: np.ndarray) -> float: |
| 23 | + """ |
| 24 | + Calculate the Mean Squared Logarithmic Error (MSLE) between two arrays. |
| 25 | +
|
| 26 | + Parameters: |
| 27 | + - y_true: The true values (ground truth). |
| 28 | + - y_pred: The predicted values. |
| 29 | +
|
| 30 | + Returns: |
| 31 | + - msle: The Mean Squared Logarithmic Error between y_true and y_pred. |
| 32 | +
|
| 33 | + Example usage: |
| 34 | + >>> true_values = np.array([1.0, 2.0, 3.0, 4.0, 5.0]) |
| 35 | + >>> predicted_values = np.array([0.8, 2.1, 2.9, 4.2, 5.2]) |
| 36 | + >>> mean_squared_logarithmic_error(true_values, predicted_values) |
| 37 | + 0.0030860877925181344 |
| 38 | + >>> true_labels = np.array([1.0, 2.0, 3.0, 4.0, 5.0]) |
| 39 | + >>> predicted_probs = np.array([0.3, 0.8, 0.9, 0.2]) |
| 40 | + >>> mean_squared_logarithmic_error(true_labels, predicted_probs) |
| 41 | + Traceback (most recent call last): |
| 42 | + ... |
| 43 | + ValueError: Input arrays must have the same length. |
| 44 | + """ |
| 45 | + if len(y_true) != len(y_pred): |
| 46 | + raise ValueError("Input arrays must have the same length.") |
| 47 | + |
| 48 | + squared_logarithmic_errors = (np.log1p(y_true) - np.log1p(y_pred)) ** 2 |
| 49 | + return np.mean(squared_logarithmic_errors) |
| 50 | + |
| 51 | + |
| 52 | +if __name__ == "__main__": |
| 53 | + import doctest |
| 54 | + |
| 55 | + doctest.testmod() |
0 commit comments