For the past couple of months, learning more about AI and data science has been on my learning backlog. However, I never actually sat down and put down a plan for learning.
The way I like to learn, is by setting myself a challenge. Just subscribing to a course, or setting the goal of learning more about X doesn’t work too well for me. I’ll share more about this around the change of the year, as I actually make 6-month-long personal development plans.
As I reflected on my development plan for these 6 months during my run this morning – running is a great way to think clearly – I noticed I needed a challenge for focusing my learning on AI and ML.
After some exploration, I stumbled across the DP-100 exam by Microsoft. It’s title “Designing and Implementing a Data Science Solution on Azure” is almost exactly what I wanted to learn more about.
One might ask, why did you decide on the DP-100 vs the AI-100. The AI-100’s title is very close to the DP-100: “Designing and Implementing an Azure AI Solution”. But comparing the learning objectives of AI-100 – which are more focused on applied AI using cognitive services and the bot framework – to the DP-100 – which are more focused on developing your own models and actual algorithms – I decided the DP-100 is closer to what I wanted to learn more about. Specifically, the learning objectives of DP-100 are:
- Define and prepare the development environment (15-20%)
- Perform feature engineering (15-20%)
- Develop models (40-45%)
In planning my learning, I did some research and decided to focus on the following learning resources:
- Modules on Microsoft Learn – focused for this exam.
- The Hundred-Page Machine Learning Book by Andriy Burkov.
- Labs on Github.
- Going through exam objectives 1-by-1 and making sure I understand them.
Modules on Microsoft Learn – focused for this exam.
The certification page included a link to Microsoft Learn, with 6 modules designed to train you for this exam.
I plan to go through all of them. Whether or not that I will reach that plan will be a question for future Nills, but it’s the plan at least.
The Hundred-Page Machine Learning Book by Andriy Burkov.
I saw this book recommended in a number of different places. From what I gather, it seems to be a perfect boil down to 150 pages of the essentials of Machine Learning. It is more theoretical than most that will be covered on the Microsoft exam, but I believe in a firm knowledge foundation.
If you believe in trying before buying as the author does, the book is available in full online. As I’ll be flying for 8 hours next week, I’ll get myself a paperback copy on Amazon – and support the author that way.
Labs on Github
I noticed these three labs on Github, specifically for the DP-100 exam. The titles themselves sound pretty interesting, focusing on actually developing models.
I’m sure there are a lot more labs available, but this seems like a good start.
Going through exam objectives 1-by-1 and making sure I understand them.
When studying for a certification exam, the way to make sure you’ll pass the exam is by going over the objectives 1 by 1 and making sure you understand the objectives. This is how I achieved the CKAD exam with a 98% score, and a process I repeat for most certification exams I study for.
The exam objectives are available here. Going through the objectives will be my last step in studying. I prefer to focus on general learning first, and only then focus on the exam. The exam in itself is a goal to force my studying, not the goal itself.
There’s a lot of learning for me to do here, and this will keep me busy for a couple of months. But I hope that by having a goal (reaching the certification) will enable me to tick of that box that says “learn about AI/ML” from my learning backlog.