Challenge-03.md

Challenge 3 – Create a Unit Test Locally and in Azure DevOps

< Previous Challenge - Home - Next Challenge >

Introduction

Use this challenge to perform tasks to confirm data has been extracted from source and saved in the working directory.

Description

• Use the code in the folder service/code/UnitTest.py to validate that AdventureWorks data is indeed downloaded and extracted into Data/ folder. Do a preview of file count in the data folder. Additionally, you could also pick a csv file visualize the data.
• TIP: Use the Python Script task in your Build pipeline to run the Unit test between tasks for TrainOnLocal.py and EvaluateModel.py.

Success criteria

• Count the number of CSV files extracted into Data/ folder using a Python script in Azure DevOps pipeline.
• Visualizing and exploring the data (using a python visualization library like matplotlib or seaborn or plotly).

Learning resources

• MLOps Home page to discover more
• MLOps documentation: Model management, deployment, and monitoring with Azure Machine Learning
• A blog on MLOps - How to accelerate DevOps with ML Lifecycle Management
• MLOps Reference Architecture