Skip to content

Model Outputs

Model outputs are typically represented as signals in Falkonry TSI, and they can be either numeric or categorical. The way these outputs are visualized and interpreted can vary based on their type and the context in which they are used. For example, a numeric model output might be visualized as a line chart showing the predicted values over time, while a categorical model output might be visualized as a bar chart showing the distribution of predicted classes.

Model outputs are the results generated by a TSI system after processing input data. These outputs can take various forms, such as predictions, classifications, explanations, synthetic signals, etc. or any other relevant information depending on the type of model and the problem it is designed to solve.

Signal Type Model Type Model Signal Type Description
User Raw (numerical or categorical) input data as received from the user.
e.g., six_machines/machine1/pressure1 
"sourceContext": { 
  ..., 
  "signalType": "USER", 
}
Model Rule Rule A boolean state indicator that represents if the rule criteria has been met or not.
e.g., machine-temp-high/rule 
"sourceContext": {
  ... 
  "modelType": "RULE",
  "modelSignalType": "RULE", 
  "signalType": "MODEL" 
}
Model Rule RuleAlert A boolean state indicator that represents if the alert criteria has been met or not.
e.g., machine-temp-high/alert 
"sourceContext": {
  ... 
  "modelType": "RULE",
  "modelSignalType": "RULEALERT", 
  "signalType": "MODEL" 
}
Model Rule Explanation A boolean state explanation that indicates if the specific signal met the rule criteria.
e.g., machine-temp-high/Explanation-six_machines/machine1/temperature3 
"sourceContext": {
  ... 
  "modelType": "RULE",
  "modelSignalType": "EXPLANATION", 
  "signalType": "MODEL" 
}
Model Calculated Calculated A calculated value (numerical or categorical) derived from the input data.
e.g., machine-temp-high/Explanation-six_machines/machine1/temperature3 
"sourceContext": {
  ... 
  "modelType": "CALCULATED",
  "modelSignalType": "CALCULATED", 
  "signalType": "MODEL" 
}
Model Anomaly AnomalyScore A numerical score > 0 indicating the magnitude of an anomaly.
e.g., six_machines/machine2/rate3/anomalyscore 
"sourceContext": {
  ... 
  "modelType": "ANOMALY",
  "modelSignalType": "ANOMALYSCORE", 
  "signalType": "MODEL" 
}
Model Condition Prediction A predicted class (categorical) label indicating the category of a condition.
e.g., machine-early-warning/predictions 
"sourceContext": {
  ... 
  "modelType": "CONDITION",
  "modelSignalType": "PREDICTION", 
  "signalType": "MODEL" 
}
Model Condition Confidence A numerical confidence score between 0 and 1 indicating the likelihood of a detected condition being true.
e.g., machine-early-warning/confidences 
"sourceContext": {
  ... 
  "modelType": "CONDITION",
  "modelSignalType": "CONFIDENCE", 
  "signalType": "MODEL" 
}
Model Condition Explanation A numerical explanation score between 1 & -1 that explains the reason for a predicted condition.
e.g., machine-early-warning/Ent/Explanation-six_amchines/machine1/temperature3 
"sourceContext": {
  ... 
  "modelType": "CONDITION",
  "modelSignalType": "EXPLANATION", 
  "signalType": "MODEL" 
}