In statistics, data science, and binary classification, a True Positive (TP) is an outcome where the model correctly predicts the positive class. It signifies that the system accurately identified the presence of a condition, hit, or event.
Together with True Negatives (TN), False Positives (FP), and False Negatives (FN), True Positives 1P) are the building blocks of the Confusion Matrix, which is used to evaluate how sensitive a test is to the presence of what it is looking for.
The Logical Breakdown
To understand a True Positive, consider the two components:
- Positive: The model predicted that the event did happen (or the condition is present).
- True: The model was correct. The event actually occurred in reality.
Simple Example:
- The Test: A medical diagnostic test for a virus.
- The Reality: The patient is actually infected with the virus.
- The Result: The test is positive.
- The Classification: This is a True Positive. The test successfully caught the condition.
Role in Performance Metrics
True Positives are the primary data point for calculating how effective a model is at finding its target:
Recall (Sensitivity)
This measures the model’s ability to find all relevant cases within a dataset. It answers: “Of everyone who actually has the condition, how many did we correctly identify?”
Loading formula...
Precision (Positive Predictive Value)
This measures how trustworthy the “Positive” results are. It answers: “Of all the times the model said ‘Positive,’ how many were actually correct?”
Loading formula...
Contextual Importance
The value of a True Positive is usually tied to the cost of missing an event:
- Medical Diagnostics: In life-threatening conditions (like heart attacks), a True Positive is critical because it triggers immediate, life-saving intervention.
- Fraud Detection: A True Positive occurs when a bank’s algorithm correctly identifies a stolen credit card transaction and freezes the account before money is lost.
- Search Engines: When you search for a specific recipe, and the first result is exactly what you wanted, that result is a True Positive for the search algorithm.
4. The Confusion Matrix
In a standard 2×2 table, the True Positive is found in the top-left cell:
| Actual: Positive | Actual: Negative | |
| Predicted: Positive | True Positive (TP) | False Positive (FP) |
| Predicted: Negative | False Negative (FN) | True Negative (TN) |
True Positives vs. False Negatives
While a True Positive is a successful detection, a False Negative (Type II Error) is a miss.
- Example: A security camera ignores a burglar because it mistakes them for a swaying tree branch.
- To increase True Positives, engineers often lower a model’s threshold, making it more sensitive. However, making a model trigger-happy to get more TPs often leads to an increase in False Positives (False Alarms).
Additional Acronyms for TP
- TP - Training Plan