Skip to content

Classification Metrics

Jump to bottom
Dominik Fuchß edited this page Aug 23, 2024 · 2 revisions

The classification metrics calculator is responsible for computing various classification performance metrics based on input classifications and ground truth data.

Input

  1. Classification: A set of classified elements.
  2. Ground Truth: A set representing the actual classification labels for comparison.
  3. String Provider Function (optional): A function that converts classification and ground truth elements into string representations for comparison purposes.
  4. Confusion Matrix Sum (optional): The sum of the confusion matrix values (true positives, false positives, etc.). Some metrics may not be calculated if this is not provided.

⚠️ Classification result entries have to match entries in the ground truth (equals)

Supported Metrics

The system calculates a variety of standard classification metrics:

  1. Precision: Measures the accuracy of the positive predictions.

    $$\text{Precision} = \frac{TP}{TP + FP}$$

    Where:

    • ( TP ) is the number of true positives.
    • ( FP ) is the number of false positives.

  2. Recall: Also known as sensitivity, recall measures the ability to find all positive instances.

    $$\text{Recall} = \frac{TP}{TP + FN}$$

    Where:

    • ( FN ) is the number of false negatives.

  3. F1-Score: A harmonic mean of precision and recall, providing a single score that balances both concerns.

    $$F1 = 2 \times \frac{\text{Precision} \times \text{Recall}}{\text{Precision} + \text{Recall}}$$

  4. Accuracy (optional): Measures the proportion of correctly predicted instances (if true negatives are provided).

    $$\text{Accuracy} = \frac{TP + TN}{TP + TN + FP + FN}$$

  5. Specificity (optional): Also called true negative rate, it measures the proportion of actual negatives that are correctly identified.

    $$\text{Specificity} = \frac{TN}{TN + FP}$$

  6. Phi Coefficient (optional): A measure of the degree of association between two binary variables.

    $$\Phi = \frac{TP \times TN - FP \times FN}{\sqrt{(TP + FP)(TP + FN)(TN + FP)(TN + FN)}}$$

  7. Phi Coefficient Max (optional): The maximum possible value for the phi coefficient.

  8. Phi Over Phi Max (optional): The ratio of the phi coefficient to its maximum possible value.

Each result includes a human-readable format that logs the computed metrics for ease of debugging and verification.

Clone this wiki locally