ClusteringSolutionVersion not predicting properly

[Community Alums]

We've been experimenting with predictive intelligence lately and exploring the it's potential regarding ITSM agents.

We are trying to utilize trained solutions for similarity, clustering and classification problems.

While working on clustering, we've trained a solution on a number of incidents. After that, we tried to predict the cluster that an incident would be added to with the following script:

var current = new GlideRecord("incident");
if (current.get("c5cb262687154e1010aa311e0ebb3565")){
    var mlSol = sn_ml.ClusteringSolutionStore.get('ml_x_emped_global_global_test_incident_cluster');

	var options = {};
	options.top_n = 1;
	options.apply_threshold = false;
	var ml_ver = mlSol.getActiveVersion();
	var results = ml_ver.predict(current, options);
	gs.info(ml_ver.getStatus(includeDetails = true));
	gs.info(results);
}

Nevertheless, the output would always be empty:

//Version Status
{"state":"solution_complete","percentComplete":"100","hasJobEnded":"true","details":{"stepLabel":"Solution Complete","percentComplete":"100","stepState":"2","stepStateLabel":"Successful","childStepDetails":[{"stepLabel":"Fetching Files for Training","stepMessage":"Downloaded all datasets (and models)","percentComplete":"100","stepState":"2","stepStateLabel":"Successful"},{"stepLabel":"Preparing Data","stepMessage":"Pipeline created","percentComplete":"100","stepState":"2","stepStateLabel":"Successful"},{"stepLabel":"Training Solution","stepMessage":"Pipeline executed successfully","percentComplete":"100","stepState":"2","stepStateLabel":"Successful"},{"stepLabel":"Uploading Trained Solution","stepMessage":"Solution uploaded successfully","percentComplete":"100","stepState":"2","stepStateLabel":"Successful"}]}}

//results
{"c5cb262687154e1010aa311e0ebb3565":[]}

we've tested on multiple clustering solutions - we never get an output.

This is on Washington Patch 1, hotfix2.

Note: we've successfully utilized classification and similarity solutions.