Recall Calculator for Machine Learning

Measure recall, sensitivity, missed positives, class averages, threshold effects, and confidence intervals while comparing classification models with clear visual results and exports for analysis.

Calculation Results

Review recall, related metrics, class results, and threshold behaviour.

Formula Substitution

Calculation Summary


            

Per-Class Recall and Supporting Metrics

Class TP FN FP TN Support Recall Precision F1 Specificity

Threshold Analysis

Threshold TP FN FP TN Recall Precision F1

Recall by Class

Threshold Performance

Calculator Inputs

Used for binary and threshold calculations.

Binary Confusion Matrix

Actual and Predicted Labels

Separate values with commas, spaces, tabs, or new lines.
The number of predicted labels must match actual labels.
Leave empty for equal weights.

CSV or Tabular Data

A header row is optional. Score enables threshold analysis.
Use zero when no score column exists.

Multilabel Classification

Enter one sample per line. Separate labels with commas.
Each line must correspond with the same actual sample.

Advanced Settings

0.50

Model Comparison

Compare recall and supporting metrics across multiple binary models.

Formula Used

Recall = TP ÷ (TP + FN)
False Negative Rate = FN ÷ (TP + FN) = 1 − Recall
Precision = TP ÷ (TP + FP)
F1 Score = 2 × Precision × Recall ÷ (Precision + Recall)

Recall measures how many actual positives the model finds. It is also called sensitivity or true positive rate. Higher recall means fewer positive cases are missed.

Macro recall gives each class equal importance. Weighted recall uses each class support as its weight. Micro recall combines all class decisions before calculation.

How to Use

  1. Select the input mode matching your available data.
  2. Enter confusion counts, labels, CSV rows, or label sets.
  3. Select the positive class and averaging method.
  4. Choose precision, zero handling, and confidence settings.
  5. Submit the form and review every reported metric.
  6. Export the summary, tables, or printable PDF report.
Recall should reflect the cost of missed positives. Medical screening often values high recall. Fraud detection may require threshold balancing.

Example Data

Scenario TP FN Recall Meaning
Medical screening95595%Five positive patients were missed.
Fraud detection703070%Thirty fraudulent cases were missed.
Defect detection1802090%Twenty defects were not detected.

Frequently Asked Questions

What does recall measure?

Recall measures detected positives among all actual positives. It focuses directly on missed positive examples. The denominator includes true positives and false negatives.

Is recall the same as sensitivity?

Yes, recall and sensitivity usually mean the same metric. True positive rate is another common name. All use identical confusion matrix values.

When is high recall especially important?

High recall matters when missed positives cause serious harm. Examples include disease detection and fraud screening. The acceptable target depends on operational consequences.

Can recall be high while precision is low?

Yes, predicting many samples as positive can raise recall. That strategy may also create false positives. Precision reveals how many positive predictions were correct.

What is macro-average recall?

Macro recall averages class recall scores equally. Rare and common classes receive equal influence. It highlights poor performance on smaller classes.

What is weighted-average recall?

Weighted recall averages class scores using class support. Larger classes influence the final value more. It reflects the observed class distribution.

How does the threshold affect recall?

Lower thresholds usually predict more positive cases. Recall often increases as false negatives decrease. Precision may fall because false positives increase.

What happens when no positives exist?

Recall becomes undefined when TP plus FN equals zero. The calculator offers configurable zero handling. Empty classes may also be ignored.

Does class imbalance affect recall?

Recall remains meaningful for each individual class. Aggregate accuracy may hide minority-class failures. Per-class recall exposes those missed positives.

Related Calculators

Confusion Matrix CalculatorClassification Accuracy CalculatorSpecificity CalculatorSensitivity CalculatorFalse Positive Rate CalculatorFalse Negative Rate CalculatorMatthews Correlation Coefficient CalculatorBalanced Accuracy CalculatorLog Loss CalculatorBinary Cross-Entropy Calculator

Important Note: All the Calculators listed in this site are for educational purpose only and we do not guarentee the accuracy of results. Please do consult with other sources as well.