What to learn if you want to become a Microsoft certified machine learning engineer
Last week I received a question through LinkedIn about the new Machine Learning certification that Microsoft offers. His question was: "Do you have some interesting resources to prepare myself for the MCSA machine learning certification". Yes, I do have those and some of them are free.
In this article I will explain what you can expect from the exam and how to prepare for the kind of questions Microsoft will ask in the exam.
Overview
The new MCSA machine learning certification is meant for people who are actively developing classic machine learning solutions on the Microsoft Platform using R. Yes, currently the certification is only available in R. Which I think is weird and should change. But alas, we'll have to deal with that limitation.
There are two exams you have to pass in order to receive the certification:
- Analyzing Big Data with Microsoft R
- Perform Cloud Data Science with Azure Machine Learning
The first exam focuses on machine learning in a on-premises environment. The second exam is focused on doing the same thing in a cloud based environment.
First exam: Analyzing Big Data with Microsoft R
The high level story behind this exam is to test your skills in the following areas:
- Loading data in R from various data sources
- Processing data in R
- Building scalable machine learning models
- Using R in different compute environments such as your local machine, a SQL Server, a compute cluster or a single remote machine
This exam revolves around Microsoft Machine Learning Server which is part of Microsoft SQL Server. So if you want to prepare for this exam I would advice you to take a look at this free online course on EdX: Analyzing Big Data with Microsoft R.
Alternatively if you live in the Netherlands you can join a classroom training at Info Support: https://training.infosupport.com/trainingen/analyzing-big-data-with-microsoft-r
Both of these courses are a great source to prepare for your exam.
Alternatively you can take a look at the various articles online. For example:
- Free DataCamp course - Introduction R for data science
- Microsoft Machine Learning Server
- RevoScaleR Package
I personally like to read a book or two on the subject and then try to build a small project that I then deploy into a "production" environment. It really helps to bridge the gap between theory and practice. This exercise works best if you don't have to focus so much on data quality etc, so grab a sample dataset to work on. Something like the iris classification dataset works great.
Second exam: Perform Cloud Data Science with Azure Machine Learning
The second exam for the MCSA machine learning certification looks very similar to the first exam. You are tested for the exact same topics, but in Azure Machine Learning Studio.
I can highly recommend the Pluralsight training Introduction to Azure Machine Learning by Jerry Kurata.
Again, if you're living in the Netherlands you can book a classroom training at Info Support: https://training.infosupport.com/trainingen/perform-cloud-data-science-with-azure-machine-learning
If you're more of an do-it-yourself person, I can recommend these sources as well:
Final thoughts
Looking at the exams and the online materials I think you can learn quite a few interesting and useful skills if you're interested in machine learning.
I have to include a disclaimer though, I haven't done any of the exams myself. I learned my data science skills through other means at for example datacamp.com and various books.
I want to thank the person who asked me the question on LinkedIn. I think I will give the exams a shot myself. It looks interesting!