Deploying custom model on Azure ML Studio

Question

In Azure ML Studio, we have the option of choosing a number of inbuilt ML models like Classification, Regression, etc. , which we can drag and drop to our workflow.

My question is, can I upload a custom ML model that I have built locally on my system in Python, and add it to the workflow?

Answer 1

1. Take the model.pkl file, zip it, and upload it into Azure Machine Learning Studio. Click the “New” icon in the bottom left:
2. In the pane that comes up, click on dataset, and then “From Local File”:
3. Select the zip file where you stored your serialized model and click the tick. You expirement should look like this:
4. Put the following code to run your classification experiment:

import pandas as pd
import sys
import pickle

def azureml_main(dataframe1 = None, dataframe2 = None):
    sys.path.insert(0,".\Script Bundle")
    model = pickle.load(open(".\Script Bundle\model.pkl", 'rb'))
    pred = model.predict(dataframe1)
    return pd.DataFrame([pred[0]])

Update If you want to declare this experiment as an API you need to add web input and output to the Python script module.