Register a Model

Before monitoring a model, it must be registered so that Domino captures information about the model through a Monitoring Config JSON. Use the JSON config to describe the model’s schema, as well as the source of the training data, and metadata about the model. For models with many features, you can include only those features that you want to monitor.

  1. From the navigation pane, click Model Monitor.

  2. From the Models page, click Register Model > Upload Model Config File.

  3. In the Register Model window, upload or paste a JSON configuration that describes the features to monitor, the location of the training dataset, and the model metadata.

    The following is a sample file. Your file must define all features that you want to monitor for drift in this configuration. These features must be specified as “variableType”: “feature”. Other variableTypes are optional when registering a model, although sample_weight and prediction_probability don’t apply. The configuration file must always include the datasetDetails and modelMetadata information.

    See Monitoring Config JSON for details about each field in the JSON.

    {
 "variables": [
   {
     "name": "age",
     "valueType": "numerical",
     "variableType": "feature"
   },
   {
     "name": "y",
     "valueType": "categorical",
     "variableType": "prediction"
   },
   {
     "name": "date",
     "valueType": "datetime",
     "variableType": "timestamp"
   },
   {
     "name": "RowId",
     "valueType": "string",
     "variableType": "row_identifier"
   }
 ],
 "datasetDetails": {
   "name": "TrainingData.csv",
   "datasetType": "file",
   "datasetConfig": {
     "path": "TrainingData.csv",
     "fileFormat": "csv"
   },
   "datasourceName": "abc-shared-bucket",
   "datasourceType": "s3"
 },
 "modelMetadata": {
   "name": "test_psg",
   "modelType": "classification",
   "version": "2",
   "description": "",
   "author": "testadmin"
 }
}

  4. Click Register Model and the model is listed on the Models page.