Learn Applied Data Science and Get Certified with Microsoft and Udacity
Unlike theoretical data science, applied data science involves additional steps to manage the lifetime of a model, which are commonly called MLOps. Azure Machine Learning is a service that conveniently supports MLOps practices, so getting to know it seems like a good idea. In this post, I will talk about the best way to learn (Udacity), and demonstrate your knowledge (Certification).
To set the expectations correctly: this exam is not so much about Data Science concepts, but rather about being able to use Azure Machine Learning service. However, when preparing for the exam, you will learn all important practices of MLOps, and also to be able to use that service effectively:
Different options for storing data inside Azure ML: Datastore vs. Dataset
Using Automated Machine Learning to train models without coding (see my earlier post on Low-Code/No-Code Machine Learning Experience)
Scheduling experiments and tracking their results (my post)
MS Learn contains a number of great courses/paths to get you introduced both to ML Concepts, and to Azure ML:
Create machine learning models learning path is a good introduction to machine learning as such. You will be using Azure ML as compute environment, but you will learn about regression, classification and deep learning.
Unlike MS Learn, taking a course at Udacity will cost you some money. However, it has some important benefits.
Udacity program is structured in such a way that upon successful completion you are guaranteed to have good knowledge of the subject, because you will not only learn the content, but also get hands-on experience working on a project. That is the reason they are ready to give certificate of completion that proves your knowledge.
During the program, you will work on three different projects. In final, capstone project, you will work on a dataset of your choice (most probably, the one from Kaggle), and perform the complete lifecycle, from importing data to getting the working model up and running in production.
You will get Azure рlayground as part of the program, which means that you would be able to experiment, and you would have cloud resources to complete the projects. If you are taking free MS Learn courses, you will need to sign up for free Azure trial yourself.
A lot of material in the course is video and not text. Many people find it easier to learn something new if they can hear other people.
Finally, during the course you will have support from instructors, as well as the community of peer learners.
The nanodegree program consists of three courses:
Azure Machine Learning Platform, which takes you through training and optimizing models using AutoML and Hyperdrive / Azure SDK. At the end of this course you will train and optimize a logistic regression model to make marketing predictions.
Machine Learning Operations course teaches you how to deploy and operationalize your models, as well as how to consume and document them. In this course you will further improve your prediction model, and deploy it to production.
Final part of the program is Capstone Project, where you can select any dataset, and build complete ML Ops pipeline around it.
While you can definitely learn Azure ML using Microsoft Learn resources, Udacity will help you organize your learning process, and immediately apply the knowledge you get in practice. The course is well aligned with Microsoft Azure Data Science Associate certification, so in addition to a certificate from Udacity, you can immediately look to get an industrial certification from Microsoft.