Accuracy score

Background

Accuracy is the simplest classification metric: the fraction of predictions that match the true labels. It is intuitive and a fine default for balanced problems, but misleading under class imbalance — a dataset that is 99% negative scores 99% accuracy by always predicting "negative". That failure mode is exactly why precision, recall, and F1 exist.

Problem statement

Implement accuracy(y_true, y_pred) returning the fraction of correct predictions:

$$\text{accuracy} = \frac{1}{n}\sum_{i=1}^{n}\mathbb{1}(y_i = \hat{y}_i)$$

Input

• y_true — array-like of labels.
• y_pred — array-like of predicted labels, the same length.

Output

Returns a float in $[0, 1]$.

Examples

Example 1

Input:  y_true = [0, 1, 1, 0, 1], y_pred = [0, 1, 0, 0, 1]
Output: 0.8

Explanation: 4 of the 5 predictions match (only index 2 differs), so accuracy $= 4/5 = 0.8$.

Constraints

• Compare elementwise and return the mean of the matches as a float.
• Works for any label type, binary or multiclass.
• The result lies in $[0, 1]$.

Notes

• Accuracy treats every sample equally, so on imbalanced data it can be high while the minority class is ignored entirely.
• It equals the trace of the confusion matrix divided by the matrix's total.