IBibm
Manage classifications models

Create classifications model

Creates a custom classifications model by uploading training data and associated metadata. The model begins the training and deploying process and is ready to use when the status is available.

post/v1/models/classifications

Request

  • Base URL
    https://api.us-south.natural-language-understanding.watson.cloud.ibm.com
  • Auth
    API key in header Authorization

Request body multipart/form-data

namestring

An optional name for the model

user_metadataobject

An optional map of metadata key-value pairs to store with this model

languagestring required

The 2-letter language code of this model

descriptionstring

An optional description of the model

model_versionstring

An optional version string

versionstring

Deprecated — use model_version

workspace_idstring nullable

ID of the Watson Knowledge Studio workspace that deployed this model to Natural Language Understanding

version_descriptionstring nullable

The description of the version

training_datastring binary required

Training data in JSON format. For more information, see Classifications training data requirements.

Example request

{
  "user_metadata": {
    "region": "North America",
    "latest": true
  }
}

Response

Model created successfully

namestring

An optional name for the model

user_metadataobject

An optional map of metadata key-value pairs to store with this model

languagestring required

The 2-letter language code of this model

descriptionstring

An optional description of the model

model_versionstring

An optional version string

versionstring

Deprecated — use model_version

workspace_idstring nullable

ID of the Watson Knowledge Studio workspace that deployed this model to Natural Language Understanding

version_descriptionstring nullable

The description of the version

featuresstring[]

The service features that are supported by the custom model

status'starting' | 'training' | 'deploying' | 'available' | 'error' | 'deleted' required

When the status is available, the model is ready to use

model_idstring required

Unique model ID

createdstring date-time required

dateTime indicating when the model was created

last_trainedstring date-time

dateTime of last successful model training

last_deployedstring date-time

dateTime of last successful model deployment

Example response

{
  "user_metadata": {
    "region": "North America",
    "latest": true
  },
  "notices": [
    {
      "message": "Training data validation failed: Too few examples for label insufficient_examples. Minimum of 5 required"
    },
    {
      "message": "Not enough examples for class 'foo'. 4 were given but 5 are required."
    },
    {
      "message": "Duplicate label given in example 27: {'foo', 'bar', 'foo'}"
    }
  ]
}

Changes

No recorded changes to this operation across all 1 revision of this API.