Predictive Intelligence : How to test Prediction and access Prediction logs - Support and Troubleshooting

Predictive Intelligence : How to test Prediction and access Prediction logs Summary Trained solution can be tested in the following ways after a Solution is trained

Instructions Test Solution Prediction

1. Test the solution using Test Solutions tab

Open the successfully trained solution which is active Scroll down and see the "Test Solution" tab Pass the input field value and top N and click "Run test"

2. Test the solution using REST API Explorer

See Test a solution prediction

3. Prediction can also be tested from background scripts. Here is a sample script

function printOutcomeArr(outcomeArr) { gs.print('################## Results ##################'); for (var i=0; i var outcome = outcomeArr[i]; gs.print((i+1) + ' : ' + outcome.predictedValue() + ', ' + outcome.predictedValueSysId() + ', ' + outcome.confidence()); } } var solutionName = 'ml_incident_categorization'; var predictor = new MLPredictor(); var solution = predictor.findActiveSolution(solutionName); var gr = new GlideRecord('incident'); gr.get('fcb4b6bedb401300864adfea5e961914'); var options={}; options.top_n = 3; options.apply_threshold = true; var outcomeArr = predictor.getPredictions(gr, solution, options); printOutcomeArr(outcomeArr); And the output would look like this

Script execution history and recovery available here

Operation Table Row Count insert ml_predictor_results_task 1 *** Script: ################## Results ################## *** Script: 1 : Email, , 47.55781164510478 *** Script: 2 : Service Catalog, , 1.6585392684522506 Related Links How to access Prediction logs 1. For the monitoring prediction results over time, see Review classification prediction results over time

2. Also prediction logs can be accessed from the localhost logs, A sample log would like this

2021-03-02 11:16:31 (016) Default-thread-100 9689951FDBAAECD43C1666041B961925 txid=3acf9113dbae *** Start #16698 /incident.do, user: admin 2021-03-02 11:16:31 (201) Default-thread-100 9689951FDBAAECD43C1666041B961925 txid=3acf9113dbae *********** Prediction Service Logs *********** 2021-03-02 11:16:31 (202) Default-thread-100 9689951FDBAAECD43C1666041B961925 txid=3acf9113dbae {"prediction_server_logs":{"id":"prediction_server_logs","logs":{"INFO":["Hostname: mlpredictor-customerxxx.xxx100.service-now.com\n"]},"allowedLogLevel":"INFO"},"{short_description\u003doutlook is not working, internal_param_predict_top_n\u003d1}":{"id":"cb12679b-2f9c-4635-a83f-56203e209d5d","logs":{},"allowedLogLevel":"INFO"}} 2021-03-02 11:16:31 (202) Default-thread-100 9689951FDBAAECD43C1666041B961925 txid=3acf9113dbae *********** Prediction Service Logs *********** 2021-03-02 11:16:31 (203) Default-thread-100 9689951FDBAAECD43C1666041B961925 txid=3acf9113dbae DxC_ML: dxc_id=898833260288746 End prediction for ml_incident_categorization took 64 ms 2021-03-02 11:16:31 (259) Default-thread-100 9689951FDBAAECD43C1666041B961925 txid=3acf9113dbae Slow business rule 'Default Task Based Prediction' on incident: INC0010011 , time was: 0:00:00.167 2021-03-02 11:16:31 (312) glide.record_watcher.evaluator.1ab8 9689951FDBAAECD43C1666041B961925 DEBUG: Executing asynchronously changeType: ENTER on responder: ResponderAction 2021-03-02 11:16:31 (355) glide.record_watcher.evaluator.1ab8 9689951FDBAAECD43C1666041B961925 [AMB] AMBQueue Published amb message, sys_id: 72cf9d17db6220183c1666041b961980 2021-03-02 11:16:31 (383) Default-thread-100 9689951FDBAAECD43C1666041B961925 txid=3acf9113dbae *** End #16698 /incident.do, user: admin, total time: 0:00:00.380, processing time: 0:00:00.380, SQL time: 0:00:00.120 (count: 58), business rule: 0:00:00.219 (count: 20), ACL time: 0:00:00.014, source: 148.139.29.229 null

3. ml_predictor_results records the results of prediction results, This can be used as a reference to see how the prediction works

4. Also all the methods described in Test Solution Prediction can be used to see the outcome of the model trained