MADEKnowledge

MADE 3.9.1 User Manual > Sensors

Sensor Parameters

Sensor Parameter Description Equations
Probability of Detection (POD) True Positive Rate. Proportion of positives that are correctly identified. The probability that a failure mode (high or low flow response) when present, will be correctly identified as such. This is calculated as the True Positives as a proportion of all positive conditions. [Image: POD.png]
Specificity (SPC) The probability that the condition being sensed is not present, given a negative reading. The probability that a nominal response when present, will be correctly identified as such. This is calculated as the True Negatives as a proportion of all negative conditions. [Image: SPC.png]
False Negative Rate (FNR) The probability that a failure mode (high or low response) when present , will be incorrectly identified as nominal. The rate of False Negatives as a proportion of all positive responses. [Image: FNR.png]
False Alarm Rate (FAR) The rate of False Positives as a proportion of all negative responses. The probability that a nominal response when present, will be incorrectly identified as a failure mode (high or low response). [Image: FAR.png]
Positive Predictive Value (PPV) The precision of the test. Calculated as the True Positives as a proportion of all positive responses. The probability a failure mode will be present when the sensor signals that a failure mode is present. [Image: PPV.png]
Negative Predictive Value (NPV) The True Negative responses as a proportion of all negative responses. The probability a nominal response will be present when the sensor signals that a nominal response is present. [Image: NPV.png]
Accuracy (ACC) The number of correct responses as a proportion of the total responses, as a percentage. The probability that the sensor will correctly identify any response (either a failure mode or nominal). [Image: ACC.png]
Balanced Accuracy (BACC) The accuracy of the test calculated in a way that avoids inflated accuracy values where the dataset is imbalanced (such as when the situation under test has a very low occurrence). Average of proportion of sensor predictions that correctly identify a failure mode and the proportion of sensor predictions that correctly identify a nominal response. [Image: BACC.png]
Likelihood Ratio Positive (LRP) The LRP is used to determine how useful the sensor is in detecting the presence of the condition. A value higher than 1 indicates that the test is associated with the condition. The higher the value, the more probable it is that the condition is present when the sensor indicates so. [Image: LRP.png]
Likelihood Ratio Negative (LRN) Likelihood Ratio Positive. The LRP is used to determine how useful the sensor is in detecting the presence of the condition. A value higher than 1 indicates that the test is associated with the condition. The higher the value, the more probable it is that the condition is present when the sensor indicates so. [Image: LRN.png]
Matthews Correlation Coefficient (MCC) The MCC provides a value which represents the predictive ability of a sensor. This value is between 1 and -1, a value of 1 represents perfect prediction, a value of -1 represents total disagreement. A value of 0 indicates a no better than random prediction. [Image: MCC.png]

Source: Local MADE 3.9.1 installation: com.phm.made.help.plugin/documents/help/html/MADeHelp/15-Sensors/15-12-SensorParameters.html · retrieved 2026-07-09