Configuring Automatic Workflows

Automatic workflows allow LicensePredictor to automatically run forecasts and evaluations for valid features after data validation. Use these settings to enable automatic workflows and configure the model, forecast horizon, and evaluation behavior.

Step 1: Open the appsettings.json file

1. Navigate to the LicensePredictor directory.

   By default, this is located at: C:\Program Files\OpeniT\LicensePredictor

2. Open the appsettings.json file.

Step 2: Configure the Automatic Workflow Settings

1. Locate the AutomaticWorkflows section in the appsettings.json file.

2. Configure the following settings:

   • Enabled — Set to true to automatically run the default forecast and evaluation for each valid feature after data validation. The default value is true.

   • ModelType — Specify the model to use for automatic forecasts and evaluations. Accepted values are FastTree, LightGBM, and MSTL. The default value is FastTree.

     Note

     If an unsupported value is specified, LicensePredictor normalizes it to a supported model.

   • ForecastHorizonYears — Specify the number of years to forecast. Accepted values are 1 through 100. The default value is 1.

     LicensePredictor converts the configured number of years into the corresponding number of cadence steps, up to 10,000 steps.

   • EvaluationPeriodYears — Specify the number of most recent years to use for evaluation. Accepted values are 1 through 100. The default value is 1.

     Recent-period evaluation is used when the feature has sufficient historical data. Otherwise, LicensePredictor uses a percentage-based evaluation split.

   • MinimumHistoryYearsForRecentEvaluation — Specify the minimum number of years of historical data required to use recent-period evaluation. Accepted values are 1 through 100. The default value is 3.

     caution

     The effective minimum history cannot be shorter than the configured EvaluationPeriodYears.

   • FallbackTrainingPercent — Specify the percentage of historical data to use for training when recent-period evaluation cannot be used. Accepted values are 10 through 90. The default value is 70.

     The remaining percentage is held out for evaluation. For example, a value of 70 uses 70% of the data for training and the remaining 30% for evaluation.

   Example: appsettings.json

   "AutomaticWorkflows": {
   "Enabled": true,
   "ModelType": "FastTree",
   "ForecastHorizonYears": 1,
   "EvaluationPeriodYears": 1,
   "MinimumHistoryYearsForRecentEvaluation": 3,
   "FallbackTrainingPercent": 70

   In this example, LicensePredictor automatically runs forecasts and evaluations using FastTree. Forecasts cover a 1-year horizon. For evaluation, LicensePredictor uses the most recent year when at least 3 years of history are available. Otherwise, it uses 70% of the available history for training and holds out the remaining 30% for evaluation.

3. Save the changes.

Step 3: Restart the LicensePredictor Service

1. Open Task Manager.

2. Under Services, restart the OpeniTLicensePredictor service.