3. Model creation

Once all data checks have passed, our modeling process kicks off automatically. This requires no action from the user. This process continuously trains, evaluates and deploys models as new data comes in.

Training a model on new data will take around 6 hours, if no unexpected issues arise. Whenever new data is uploaded, the model will be retrained and redeployed automatically.

To check whether the model for a ship has been deployed, query the Model Metadata endpoint. This will return an HTTP 404 Not Found error as long as the model is not online.

In conclusion, model creation requires no actions from the user. It happens automatically after the required data is uploaded, all data checks have passed, and the model passes Toqua's internal validation checks.

What’s Next