templates.inference.inference.impl.py.vm Maven / Gradle / Ivy

Go to download
"""
Implementation of the inference analytic.

GENERATED STUB CODE - PLEASE ***DO*** MODIFY

Originally generated from: ${templateName}
"""
import logging

from ${artifactIdPythonCase}.config.inference_config import InferenceConfig
from ${artifactIdPythonCase}.validation.inference_message_definition import RequestBody, ResponseBody
from ${artifactIdPythonCase}.validation.inference_payload_definition import Inference

model = None
def load_model():
    """
    Loads the trained model based on the configuration defined in InferenceConfig and make it available for
    utilization for inferencing.
    :return:
    """
    config = InferenceConfig()

    try:
        # Load the model from the model directory
        # model = mlflow.sklearn.load_model(config.model_directory())
        
        pass 
    except Exception:
        logging.exception(f'Failed to load model at {config.model_directory()}')

def execute_inference(request: RequestBody):

    # Prep the data from the inference request
    prepped_data = request.prep_data()


    # Use the model to predict using the prepped data
    # if not model:
    #     load_model()
    #     predictions = model.predict(prepped_data)

    # Process single and batch requests as appropriate
    if len(request.data) > 1:
        pass
    else:
        pass

    return ResponseBody(inferences=[Inference(prediction="single-prediction", score=10)])