MLPredictor - Global (deprecated)

The MLPredictor API provides utility methods for Predictive Intelligence predictions.

Note:

This API has been deprecated and is intended to be removed in a future release. Refer to Using ML APIs for the most recent guidelines.

The MLPredictor API requires the Predictive Intelligence plugin (com.glide.platform_ml) and is provided within the sn_ml namespace.

This class contains all methods necessary to get prediction results from input data.

MLPredictor - MLPredictor()

Instantiates a new MLPredictor object.

Table 1. Parameters

Name	Type	Description
None

The following example shows how to use an MLPredictor object in a business rule to record final values after a prediction.

function executeRule(current, previous /*null when async*/) {
	var predictor = new MLPredictor();
	predictor.recordFinalValuesInPredictionResults(current, "On close");

}(current, previous);

MLPredictor - applyPrediction(GlideRecord now_GR, Array solutions)

Sets predicted values from an array of specified solutions to a specified record.

Table 2. Parameters

Name	Type	Description
now_GR	GlideRecord	The record on which to apply the array of predicted solutions.
solutions	Array	Specified solution objects associated with the GlideRecord.

Table 3. Returns

Type	Description
void

The following example demonstrates using the applyPrediction() method in a business rule.

(function executeRule(current, previous /*null when async*/) {
    var mlPredictor = new MLPredictor();
    //Get the list of active solutions for the glide record table
    var solutions = mlPredictor.findActiveSolutionsForRecord(current);
    //Run prediction and apply predicted value to the record
    mlPredictor.applyPrediction(current, solution);

})(current, previous);

MLPredictor - applyPredictionForSolution(GlideRecord now_GR, Object solution)

Applies a predicted value from a specified classification solution to the specified GlideRecord.

For each solution in the GlideRecord, call this method to predict the results and set the field value on the incident to those results.

Table 4. Parameters

Name	Type	Description
now_GR	GlideRecord	GlideRecord object containing values on which to run a prediction and apply the results.
solution	Object	Classification solution object to be executed.

Table 5. Returns

Type	Description
Boolean	True upon prediction success, error otherwise.

To use this template, copy the Incident Based Prediction (Template) business rule, make the new record Active, and follow the commented instructions to initialize the solutionNames variable.

(function executeRule(current, previous /*null when async*/) {
	var solutionNames = ["solution1", "solution2", ...];

	/* For domain separation (MSP) use case */
//	var solutionNames;
//	if (current.sys_domain == "A")
//		solutionNames = ["solution_A1", "solution_A2", ...];
//	else if (current.sys_domain == "B")
//		solutionNames = ["solution_B1", "solution_B2", ...];
//	else
//		...

	var predictor = new MLPredictor();
	var info = "";
	solutionNames.forEach(function(solutionName) {
		var solution = predictor.findActiveSolution(solutionName);
		if (!solution)
			return;

		/* The next line of code does the prediction and updates the current record. */
		predictor.applyPredictionForSolution(current, solution);

		/* If no prediction is done, do not build the prediction info message. */
		if (!predictor.applyPredictionForSolution(current, solution))
			return;

		/* If user doesn't have 'itil' role, we don't build prediction info message. */
		if (!gs.hasRole('itil'))
			return;

		/* Building prediction info message */
		var fieldName = solution.getPredictedField();
		var fieldLabel = current.getElement(fieldName).getED().getLabel();
		var predictedDisplayValue = current.getDisplayValue(fieldName);
		var msg = gs.getMessage("Predicted {0} for {1}.", [predictedDisplayValue, fieldLabel]);
		if (info.length > 0)
			info += " ";
		info += msg;
	});
	/* Print out prediction info message on screen. */
	if (info.length > 0) {
		var incidentUrl = "<a href='"+current.getLink()+"'>"+current.number+":</a>";
		gs.addInfoMessage(incidentUrl + " " + info);
	}
})(current, previous);

MLPredictor - findActiveSolution(String solutionName)

Gets the solution object.

This method only returns the solutions if the ml_solution definition and solution are active.

Table 6. Parameters

Name	Type	Description
solutionName	String	Name of the ml_solution record.

Table 7. Returns

Type	Description
Object	Solution object for the specified solutionName if the ml_solution definition and solution is active, null otherwise.

This example takes a hard-coded value and passes it to a specified Machine Learning model. You can use the outcome and confidence variables to set values or other business logic.

var solutionName = 'ml_incident_assignment';
  var shortDescriptionValue = "Unable to connect!"
  var input = {
  short_description : shortDescriptionValue
  };
  var MLP = new MLPredictor();
  var solution = MLP.findActiveSolution(solutionName);
  var predictedOutcome = solution.predictText(input);
  var outcome = predictedOutcome.predictedValue();
  var confidence = predictedOutcome.confidence();
  gs.info("Predicted value: " + outcome)
  gs.info("Confidence: " + confidence)
        

Output:

DxC_ML:
        ******* Finished Prediction********* *** Script: Predicted value: Software *** Script:
        Confidence: 80.78079080096245

MLPredictor - findActiveSolutionsForRecord(GlideRecord now_GR)

Gets active solutions for a table in a specified GlideRecord.

Table 8. Parameters

Name	Type	Description
now_GR	GlideRecord	GlideRecord from which to collect active solution objects.

Table 9. Returns

Type	Description
Array	Array of active solution objects associated with the table that the specified record is for.

This script passes a GlideRecord (such as an incident), and processes all solutions for the GlideRecord, returning the values for each.

/* This is only to get a hard-coded GR */
     var current = new GlideRecord('incident');
     current.get('965c9e5347c12200e0ef563dbb9a7156');
     var predictor = new MLPredictor();
     var solutions = predictor.findActiveSolutionsForRecord(current);
     solutions.forEach(function(solution) {
         var outcome = solution.predict(current);
         /* Use this to set the field to the predicted value in the GlideRecord */
         var fieldName = solution.getPredictedField();
         current[fieldName] = outcome.predictedValue();
         current.update();

         gs.info("Predicted value: " + outcome.predictedValue())
         gs.info("Confidence: " + outcome.confidence())
});

Output (Assignment Group prediction):

DxC_ML: ******* Finished Prediction*********
        *** Script: Predicted value: Software *** Script: Confidence:
        99.74530683715315

Output (Category prediction):

DxC_ML:
        ******* Finished Prediction********* *** Script: Predicted value: Software *** Script:
        Confidence: 98.7172805135524

MLPredictor - getPredictedValue(Object solution, Object outcome)

Gets the predicted value for a specified solution based on the specified prediction outcome.

Table 10. Parameters

Name	Type	Description
solution	Object	Solution from which to get the predicted value.
outcome	Object	Prediction outcome results for the specified solution (var outcome = solution.predict(now_GR)).

Table 11. Returns

Type	Description
String	Predicted value for specified solution based on the specified outcome of the prediction.
Null	Returns null if the prediction confidence does not satisfy thresholds.

MLPredictor - getPredictions(GlideRecord now_GR, Object solution, Object options)

Gets predictions for a specified solution.

Table 12. Parameters

Name	Type	Description
now_GR	GlideRecord	GlideRecord to be predicted.
solution	Object	Solution object to be executed.
options	Object	Optional JSON object key value pair with the following properties: options.top_n: If provided, returns results up to the expected number of predictions, otherwise default is read from the glide.platform_ml.max_num_predictions system property. options.apply_threshold: Checks the threshold value (solution threshold for similarity, class level threshold for classification) for the solution and applies it to the result set. Default value is true.

Table 13. Returns

Type	Description
Array	Array of predicted outcome objects

The following example calls a specified solution and executes predictions on it.

function printOutcomeArr(outcomeArr) {
gs.print('################## Results ##################');
for (var i=0; i<outcomeArr.length; i++) {
    var outcome = outcomeArr[i];
    gs.print((i+1) + ' : ' + outcome.predictedValue() + ', ' + outcome.predictedValueSysId() + ', ' + outcome.confidence());
}
}

var solutionName = 'ml_x_snc_global_prop_flip_test';
var predictor = new MLPredictor();
var solution = predictor.findActiveSolution(solutionName);

var now_GR = new GlideRecord('incident');
now_GR.get('1c741bd70b2322007518478d83673af3');

var options = {};
options.top_n = '10';            // top_n is an integer between 1 and 1000
options.apply_threshold = false; // Value can be set to true or false

printOutcomeArr(predictor.getPredictions(now_GR, solution, options));

MLPredictor - isClassificationSolution(Object solution)

Identifies if a solution object is a classification type.

Table 14. Parameters

Name	Type	Description
solution	Object	Name of the ML solution.

Table 15. Returns

Type	Description
Boolean	Returns true if the input solution is a classification type, false otherwise.

var isClassificationSolution = this.isClassificationSolution(solution);
 
	//classification solution each class has different threshold
	//therefore needs to get all the results from ml engine
	if (applyThreshold && isClassificationSolution) {
		var maxClassificationTopN = 50;
		outcomeArr = solution.predictTopN(now_GR, maxClassificationTopN);
	}
	else  {
	outcomeArr = solution.predictTopN(now_GR, topN);
	}
 
	if (outcomeArr === null) {
		//instead of returning null returning empty array
		return [];
	}

MLPredictor - isSimilaritySolution(Object solution)

Identifies if a solution object is a similarity type.

Table 16. Parameters

Name	Type	Description
solution	Object	Name of the ML solution; for example, ml_incident_categorization.

Table 17. Returns

Type	Description
Boolean	Returns true if the input solution is a similarity type, false otherwise.

MLPredictor - outcome.confidence()

Gets the confidence of the predicted value.

Table 18. Parameters

Name	Type	Description
None

Table 19. Returns

Type	Description
String	The estimated precision of the prediction as a percentage. For example, 53.84615375762915.

var MLP = new MLPredictor();
  var solution = MLP.findActiveSolution(solutionName);
  var predictedOutcome = solution.predictText(input);
  var outcome = predictedOutcome.predictedValue();
  var confidence = predictedOutcome.confidence();
  gs.info("Predicted value: " + outcome)
  gs.info("Confidence: " + confidence)

MLPredictor - outcome.predictedValue()

Gets the predicted value from the MLPredictor outcome object.

Table 20. Parameters

Name	Type	Description
None

Table 21. Returns

Type	Description
String	Predicted value from the Outcome object.

var MLP = new MLPredictor();
  var solution = MLP.findActiveSolution(solutionName);
  var predictedOutcome = solution.predictText(input);
  var outcome = predictedOutcome.predictedValue();
  var confidence = predictedOutcome.confidence();
  gs.info("Predicted value: " + outcome)
  gs.info("Confidence: " + confidence)

MLPredictor - outcome.predictedValueSysId()

Gets the sys_id of the predicted value.

Table 22. Parameters

Name	Type	Description
None

Table 23. Returns

Type	Description
String	Predicted value sys_id.

function printOutcomeArr(outcomeArr) {
gs.print('################## Results ##################');
for (var i=0; i<outcomeArr.length; i++) {
    var outcome = outcomeArr[i];
    gs.print((i+1) + ' : ' + outcome.predictedValue() + ', ' + outcome.predictedValueSysId() + ', ' + outcome.confidence());
}

MLPredictor - recordFinalValuesInPredictionResults(GlideRecord now_GR, String reason)

Sets final prediction result values to a specified GlideRecord with an optionally specified reason.

Table 24. Parameters

Name	Type	Description
now_GR	GlideRecord	GlideRecord on which to set the final prediction result values.
reason	String	Optional. Reason for applying results.

Table 25. Returns

Type	Description
void

In following example, the recordFinalValuesInPredictionResults() method is called when the incident is closed.

(function executeRule(current, previous /*null when async*/) {
	var predictor = new MLPredictor();
	predictor.recordFinalValuesInPredictionResults(current, "On close");

})(current, previous);

MLPredictor - solution.getCapability()

Gets the capability information of a trained solution.

Table 26. Parameters

Name	Type	Description
None

Table 27. Returns

Type	Description
String	Definition ID and version of the trained solution, error message otherwise

MLPredictor - solution.getName()

Gets the name of the solution used for prediction.

Table 28. Parameters

Name	Type	Description
None

Table 29. Returns

Type	Description
String	The name of the solution to use for predictions; for example, ml_incident_categorization

MLPredictor - solution.getPredictedField()

Gets the predicted value of a solution.

Table 30. Parameters

Name	Type	Description
None

Table 31. Returns

Type	Description
String	Value of a solution's predicted output field

/* Get a hard-coded GR */
     var current = new GlideRecord('incident');
     current.get('965c9e5347c12200e0ef563dbb9a7156');
     var predictor = new MLPredictor();
     var solutions = predictor.findActiveSolutionsForRecord(current);
     solutions.forEach(function(solution) {
         var outcome = solution.predict(current);
         /* Use this to set the field to the predicted value in the GlideRecord */

         var fieldName = solution.getPredictedField();
         current[fieldName] = outcome.predictedValue();
         current.update();
         gs.info("Predicted value: " + outcome.predictedValue())
         gs.info("Confidence: " + outcome.confidence())
});

MLPredictor - solution.getThreshold(String className)

Gets the solution threshold.

The threshold represents a percentage reflecting the minimum prediction accuracy.

Table 32. Parameters

Name	Type	Description
className	String	A specified categorical value of the solution output field

Table 33. Returns

Type	Description
Number	Value of the threshold represented as a percentage between 0 and 100.

MLPredictor - solution.getVersion()

Gets the version of the active solution.

Table 34. Parameters

Name	Type	Description
None

Table 35. Returns

Type	Description
String	Version of the active solution

MLPredictor - solution.isActive()

Determines if the specified solution is active.

Table 36. Parameters

Name	Type	Description
None

Table 37. Returns

Type	Description
Boolean	True if the solution is active, false otherwise

MLPredictor - solution.predict(GlideRecord now_GR, Object threshold)

Gets solution prediction results as an Outcome object.

Table 38. Parameters

Name	Type	Description
now_GR	GlideRecord	GlideRecord of the solution input table
threshold	Object	Threshold value (solution level threshold for similarity, class level threshold for classification)

Table 39. Returns

Type	Description
Object	Prediction outcome result of the specified solution (var outcome = solution.predict(now_GR))

solutions.forEach(function(solution) {
         var outcome = solution.predict(current);
         /* Use this to set the field to the predicted value in the GlideRecord
         var fieldName = solution.getPredictedField();
         current[fieldName] = outcome.predictedValue();
         current.update();
         */
         gs.info("Predicted value: " + outcome.predictedValue())
         gs.info("Confidence: " + outcome.confidence())
});

MLPredictor - solution.predictTopN(GlideRecord now_GR, Object topN)

Returns a list of outcome objects up to the expected number of predictions. Maximum number 1000 predictions.

Table 40. Parameters

Name	Type	Description
now_GR	GlideRecord	GlideRecord of the solution
topN	Object	Expected number of predictions, any number over 1000 returns 1000 results

Table 41. Returns

Type	Description
Array	List of outcome objects in an array including GlideRecord, threshold, system ID, and expected number of predictions (topN object)

var isClassificationSolution = this.isClassificationSolution(solution);
 
	//classification solution each class has different threshold
	//therefore needs to get all the results from ml engine
	if (applyThreshold && isClassificationSolution) {
		var maxClassificationTopN = 50;
		outcomeArr = solution.predictTopN(now_GR, maxClassificationTopN);
	}
	else  {
		outcomeArr = solution.predictTopN(now_GR, topN);
	}
 
	if (outcomeArr === null) {
		//instead of returning null returning empty array
		return [];
	}