Prediction from Models

DD platform allows to use models for prediction, i.e. pass data get output such as image labels or text classification either:

• Via User Interface

• Via API with Python, Javascript, Shell, or any other client

Prediction from UI

• Model setup, simply click on the ‘Add Service’ button from a model badge in the list on the Predict page:

Follow the steps below:

• Click on Predict to reach the model section

• Locate your model and click ‘Add Service’

• Two possibilities from here:

  • The model loads automatically and you are ready to use it through the UI

  • The model needs to be setup and you need to fill the parameters up, see the screenshot on the right

• One a model is loaded up, you can either pass data as URL or using data you have uploaded, with the ‘Path’ option

Prediction from API

Python, Javascript and Shell code available in the UI

The Python, Javascript and Shell code are made available for every model call directly from the UI. This makes it easy to Copy & Paste the code to your application.

The JSON tab yields the full model output in JSON format, exactly as would be received by your application.

Available clients:

• Python client, also available from the main works DD platform Jupyter directory.

  • Put the dd_client.py file into your project directory

• Javascript client

• See the ecosystem page for more clients and libraries.