Skip to content

statsim/confusionmatrix

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 
 
 
 
 

Repository files navigation

Generate Confusion Matrix and Evaluation Metrics Online

A confusion matrix is a useful tool for evaluating the performance of classification models. It provides a breakdown of predicted versus actual outcomes, allowing for a deeper understanding of model performance beyond just accuracy.


Why Accuracy is Not Enough for Binary Classification?

Accuracy is the proportion of correctly classified instances among all instances:

```latex \text{Accuracy} = \frac{\text{TP} + \text{TN}}{\text{Total}} ```

Where $$\text{TP}$$ is the number of true positives, $$\text{TN}$$ is the number of true negatives, and $$\text{Total}$$ is the total number of instances.

Limitations of Accuracy as a Metric

  • Class Imbalance. In datasets with imbalanced classes (e.g., 95% negatives and 5% positives), a model that always predicts the majority class (negative) can achieve high accuracy but fails to capture the minority class.
  • No Insight into Error Types. Accuracy does not distinguish between types of errors (e.g., false positives vs. false negatives), which can have vastly different implications in real-world scenarios.

Confusion Matrix Terminology

Predicted Positive Predicted Negative
Actual Positive True Positive (TP): Correctly identified positive cases False Negative (FN): Missed positive cases
Actual Negative False Positive (FP): Incorrectly predicted positive cases True Negative (TN): Correctly identified negative cases

Key Terms:

  • TP (True Positive): Correctly predicted positive instances.
  • TN (True Negative): Correctly predicted negative instances.
  • FP (False Positive): Negative instances incorrectly predicted as positive.
  • FN (False Negative): Positive instances incorrectly predicted as negative.

Metrics and Their Meaning

1. Sensitivity (True Positive Rate, TPR)

```latex \text{TPR} = \frac{\text{TP}}{\text{TP} + \text{FN}} ```
  • Definition: Proportion of actual positives correctly identified.
  • Importance: Measures the ability to detect positive cases (useful in medical diagnosis, fraud detection).

2. Specificity (SPC)

```latex \text{SPC} = \frac{\text{TN}}{\text{FP} + \text{TN}} ```
  • Definition: Proportion of actual negatives correctly identified.
  • Importance: Measures the ability to avoid false alarms (useful in spam filters, anomaly detection).

3. Precision (Positive Predictive Value, PPV)

```latex \text{PPV} = \frac{\text{TP}}{\text{TP} + \text{FP}} ```
  • Definition: Proportion of positive predictions that are correct.
  • Importance: Indicates reliability of positive predictions.

4. Negative Predictive Value (NPV)

```latex \text{NPV} = \frac{\text{TN}}{\text{TN} + \text{FN}} ```
  • Definition: Proportion of negative predictions that are correct.
  • Importance: Indicates reliability of negative predictions.

5. False Positive Rate (FPR)

```latex \text{FPR} = \frac{\text{FP}}{\text{FP} + \text{TN}} ```
  • Definition: Proportion of actual negatives incorrectly predicted as positive.
  • Importance: Highlights the rate of false alarms.

6. False Discovery Rate (FDR)

```latex \text{FDR} = \frac{\text{FP}}{\text{FP} + \text{TP}} ```
  • Definition: Proportion of positive predictions that are incorrect.
  • Importance: Complements precision in evaluating prediction quality.

7. False Negative Rate (FNR)

```latex \text{FNR} = \frac{\text{FN}}{\text{FN} + \text{TP}} ```
  • Definition: Proportion of actual positives incorrectly predicted as negative.
  • Importance: Highlights the rate of missed positive cases.

8. F1 Score

```latex \text{F1} = \frac{2 \cdot \text{TP}}{2 \cdot \text{TP} + \text{FP} + \text{FN}} ```
  • Definition: Harmonic mean of precision and recall.
  • Importance: Balances precision and recall, especially useful in imbalanced datasets.

9. Accuracy

```latex \text{ACC} = \frac{\text{TP} + \text{TN}}{\text{P} + \text{N}} ```
  • Definition: Proportion of correct predictions.
  • Importance: Provides a general sense of model performance but is less reliable in imbalanced datasets.

10. Matthews Correlation Coefficient (MCC)

```latex \text{MCC} = \frac{\text{TP} \cdot \text{TN} - \text{FP} \cdot \text{FN}}{\sqrt{(\text{TP} + \text{FP})(\text{TP} + \text{FN})(\text{TN} + \text{FP})(\text{TN} + \text{FN})}} ```
  • Definition: A balanced measure that accounts for TP, TN, FP, and FN.
  • Importance: Considered a robust metric for imbalanced datasets. Ranges from -1 (inverse prediction) to +1 (perfect prediction), with 0 indicating random performance.

Key Takeaways

  1. Class Imbalance Requires Caution: Accuracy alone can be misleading when classes are imbalanced.
  2. Use Multiple Metrics: Evaluate sensitivity, specificity, precision, and other metrics to understand the trade-offs in your model.
  3. MCC for Imbalanced Datasets: Use Matthews Correlation Coefficient for a single comprehensive measure.
  4. Domain-Specific Importance: Choose metrics based on the problem domain (e.g., sensitivity for medical tests, precision for legal applications).

By understanding and applying these metrics, you can better assess and improve your model's performance, ensuring it meets the requirements of the task at hand.