|
TERMS |
- True Positive = correct estimation : Actual TRUE condition is correctly estimated as TRUE by algorithm.
- True Negative = correct estimation : Actual FALSE condition is correctly estimated as FALSE by algorithm.
- False Positive = incorrect estimation : Actual FALSE condition is incorrectly estimated as TRUE by algorithm.
- False Negative = incorrect estimation : Actual TRUE condition is incorrectly estimated as FALSE by algorithm.
-------------------
- Accuracy = ability to estimate correctly in actual BOTH conditions.
- Sensitivity = ability to estimate correctly in actual TRUE condition.
- False Negative Rate = possibility of Actual TRUE samples which are unable to find.
- Specificity = ability to estimate correctly in actual FALSE condition.
- False Positive Rate = possibility of Actual FALSE samples which are unable to find.
- Positive Predictability = confidence of positive estimation.
- Negative Predictability = confidence of negative estimation.
|
DEFINITIONS |
- Accuracy = ( True Positive + True Negative ) / Number of all samples
- Sensitivity = ( True Positive ) / Number of Actual TRUE samples
- False Negative Rate = ( Unfound Number of Actual TRUE samples ) / Number of Actual TRUE samples
- Specificity = ( True Negative ) / Number of Actual FALSE samples
- False Positive Rate = ( Unfound Number of Actual FALSE samples ) / Number of Actual FALSE samples
- Positive Predictability = ( True Positive ) / ( True Positive + False Positive )
- Negative Predictability = ( True Negative ) / ( True Negative + False Negative )
-------------------
- False Negative Rate + Sensitivity = 1
- False Positive Rate + Specificity = 1
|
DIAGRAMMATIC VIEW (rewrite from Wikipedia) |
|
Actual Condition |
|
True |
False |
Algorithmic Estimation |
Positive |
True Positive |
False Positive |
Positive Predictability |
Negative |
False Negative |
True Negative |
Negative Predictability |
|
Sensitivity |
Specificity |
Accuracy |
|
REFERENCES |
|