Execute Custom ML Endpoint

The code snippet given below allows you to pass input data to a published QuickML endpoint , and get the inferences from the ML model . The output returns the prediction of the values of the target column that is defined while creating the ML pipeline.

The quickml component instance is created as shown, which will not fire a server-side call. You will need to create a data dictionary through which you can pass the input data to the model’s endpoint as key-value pairs. The dictionary keys must match the features from your trained dataset that is expected by the model. The endpoint_key mentioned is the unique ID of the endpoint published for the ML model configured in your project. The endpoint key and the input data are passed to the run_inference ( endpoint_key, input_data) method for execution. The app reference used in the code below is the Python object returned as a response during SDK initialization.
Sample Code Snippet

# Create a QuickML instance.
quickml = app.quick_ml()
# Run Inference

# Replace with your endpoint key copied from the Catalyst console.
endpoint_key = "<ENDPOINT_KEY>"

# Replace the sample feature names and values with the input expected by your model.

input_data = {
“<FEATURE_1>”: “<VALUE_1>”,
“<FEATURE_2>”: “<VALUE_2>”
}
response = quickml.run_inference (endpoint_key, input_data)
print(response)

The syntax of the model response received is shown below:

{
   "status": "success",
  "result": [
    {
      "prediction": "1",
      "confidence": 0.87
    }
  ]
}

Parameters Used

Where to find the endpoint information?
Create an endpoint for your trained ML model and access the endpoint details page to view the Endpoint URL, required headers and a sample request response.