Hands on experience with Azure Data Scientist Certification
In this article, I would like to give some recommendations on how to prepare for the Azure Data Scientist Certification from Microsoft (DP-100) based on my personal experience.
To pass, you have to earn 700 out of 1000 (70 %) points. This exam is relatively difficult. Those who have done several Microsoft certifications say that was one of the most challenging. So you have to be well prepared.
During the exam, you have 100 minutes and 45–60 questions. If there are labs, then you have an additional 20 minutes (120 minutes total). There have been no labs in DP-100 in recent years, but they should always be expected.
It is expected that you are already familiar with the Python programming language and the basics of machine learning. So the main focus in preparation is the Azure Machine Learning platform and SDK (not the basic theory of machine learning).
It is important to understand that there is no official complete preparation single course, that you can study in order to cover 100% of the topics that will be in the exam.
Each exam is slightly different and can contain completely new technical detailed questions about some technologies, libraries, or Microsoft products that were not covered by trainings and are described only in official documentation.
Before taking this exam, please check the actual topics list (DP-100 study guide) from the official page:
Below is a recommended minimum:
A CLOUD GURU
At first, it is recommended to watch all videos and do all labs from the next course in A Cloud Guru which has 3 parts:
- DP-100 Part 1 - Preparation
- DP-100 Part 2 - Modeling
- DP-100 Part 3 - Deployment and Working with SDK
At the end of Part 3 is a practical exam, but it is several times easier than the real Microsoft exam. It has only about 50 questions and doesn’t cover even 50% of the topics from the actual exam. So don’t rely too much on this course and this practical test exam, you should definitely study more information to be well prepared. The course leader says that you don’t need a deep understanding of the Azure SDK, but the actual exam has plenty of questions about it (in Python).
LEARNING PATHS FROM MICROSOFTExam DP-100: Designing and Implementing a Data Science Solution on Azure - Certifications
On the official DP-100 exam page, read carefully the information from the next learning paths at the bottom:
Create machine learning models - Training
This section can be skipped if you are already familiar with Python and all the basic algorithms of Machine Learning. If you have no knowledge of Machine Learning, then it is better to read additional information somewhere else because this section is too short.
Microsoft Azure AI Fundamentals: Explore visual tools for machine
learning - Training
This is optional, you should be already familiar with ML Azure Studio if you have done all labs from the A Cloud Guru course above.
EBuild and operate machine learning solutions with Azure Machine
Learning - Training
This is the most important section! It is about the Azure SDK. Most of the questions in the exam are connected with this section! So it is highly recommended to read it carefully several times and pay additional attention to:
- names of imported libraries;
- function names and packages;
- names and order of function arguments;
- order of function calls;
- order of the main steps when we make deployments and build pipelines;
- memorize the main commands in the Azure Command-Line Interface (CLI).
Machine Learning with Azure Databricks - Training
Usually exam doesn’t have questions about Azure Databricks, but several of them still could be from this topic. So it could be useful to read this section if you want to be better prepared.
- Create machine learning models - Training
COURSERA Microsoft Azure Data Scientist Associate (DP-100) Professional CertificateMicrosoft Azure Data Scientist Associate (DP-100)
It is recommended to complete also the official preparation course from Microsoft on the Coursera platform, but it is not obligatory. Many topics were already described in the learning paths above. Although these 5 courses also don’t cover all questions from the exam, they could help fill in some of your knowledge gaps.
Don’t pay attention to the 7-month course duration; you could watch just all videos and answer all questions in quizzes without labs in several days. You can accomplish each course without labs in approximately 1 day erledigen (although it is written 6 weeks).
You have a 7-day free trial period. After this time, you cannot submit final quizzes and earn certificates. Intermediate quizzes will still work after 7 days, and you will have further access to all videos and all texts.
These courses are built around the same topics that were described in the previous section (learning paths from the official DP-100 exam page). Some topics are covered in Coursera in more detail; others were described in more detail in learning paths before.
This Coursera specialization contains 5 courses:
Create Machine Learning Models in Microsoft Azure
This course is very short and optional; it could be skipped because you should be already familiar with these topics.
Microsoft Azure Machine Learning for Data Scientists
This course is also optional and could be skipped. If you have done labs in A Cloud Guru, then you should be already familiar with these topics.
Build and Operate Machine Learning Solutions with Azure
This course is the most important! It is directly connected with Build and operate machine learning solutions with Azure Machine Learning - Training (Azure SDK) from the previous learning path section, although in the learning path above it was described in more detail.
Perform Data Science with Azure Databricks
This course is connected with Machine learning with Azure Databricks - Training from learning path above. As it was already mentioned, exam usually don’t have questions about Databricks, but still sometimes there are questions from this topic. So it is recommended at least to finish all Quizzes from this course. This course contains only short information and is not very well explained (low rating of 3.5 out of 5.0 stars). Maybe it is better to read just learning paths.
Prepare for DP-100: Data Science on Microsoft Azure Exam
This course is actually just a test practical exam with some recommendations on how to prepare for it. But please keep in mind that even the official practical test from Microsoft on the Coursera platform is much easier than the actual exam! If you always answer 80% or more questions in all these quizzes, it still doesn’t mean that you can easily achieve 80% or even 70% (the required minimum) in a real exam. So you should do more preparation using questions similar to those in real Microsoft exams as described below (e.g., Mindhub).
If you finish all 5 courses (is enough just to finish all quizzes, and maybe you should click on the end of each video to get a green check mark), then you ge 5 Coursera certificates and a 50% discount for the official DP-100 exam.
- Create Machine Learning Models in Microsoft Azure
AZURE MACHINE LEARNING OFFICIAL DOCUMENTATION FROM MICROSOFT:Azure Machine Learning documentation
It is highly recommended to read at least some sections of the official documentation from Microsoft. Which concrete sections you should read will be more clear when you start to do practical tests further. All questions usually have references to them. Exam questions can sometimes be very specific and are explained only in this official documentation. I would not recommend spending all your time reading this documentation; it is enough to read those topics that were not explained in the trainings from A Cloud Guru and Coursera above.
It is highly recommended to read at least the Concepts section.
PRACTICAL TEST FROM MINDHUB (RECOMMENDED BY MICROSOFT):DP-100: Designing and Implemneting a Data Science Solution on Azure Microsoft Certification Practice Test
This test contains 130 unique questions and is officially recommended by Microsoft for exam preparation. The cost is 90–100 USD plus value-added tax (19% in Germany). You can find additional discount voucher codes on the internet (about a 25% or 30% discount). You should keep in mind that these questions are not met in exams; you can do them just for practice to understand which topics you should study additionally.
PRACTICAL TEST FROM WHIZLABS (IS SPONSORED BY MICROSOFT):Microsoft Azure Exam DP-100 Certification - Whizlabs
This test contains 136 unique questions. It is sponsored by Microsoft. The cost is 20 USD, but you can easily find a discount voucher code for a 30% discount on the internet. None of these questions will be met in real exams, but you can use them for additional preparation. These questions are easier than the ones from Mindhub above.
EXAMPLES OF SOURCE CODE FROM MICROSOFT (OPTIONAL):mslearn-dp100
MachineLearningNotebooks/how-to-use-azureml at master . Azure/MachineLearningNotebooks
This section is completely optional. Usually, it is enough to learn the Azure SDK and do practical tests, which contain a lot of questions about source code.
CHAT GPT (OPTIONAL):Chat GPT
You can also use Chat GPT as a free coach who can answer almost any specific, difficult question that was not clearly explained during your studies. Chat GPT answers correctly to most questions from the DP-100 exam (in 80%–85% of cases), although sometimes it makes errors. There is no warranty that all the answers are correct! But in my case, Chat GPT still helped me a lot to prepare better for the exam. It has clarified the answers to some really specific questions. It was not always easy to find them on the internet. So, I would highly recommend this chatbot for preparations, although it is optional.
If you are very limited in time and cannot study all the topics above, then here is the absolute minimum:
- A Cloud Guru;
- Learning path from Microsoft (Build and operate machine learning solutions with Azure Machine Learning);
- Practical tests.
Thanks for reading this article! I hope my recommendations will help you prepare better for this certification.