From Zero to Hero: how to learn Power BI quickly - excerpts from my learning journey.

“Learning never exhausts the mind.” – Leonardo da Vinci

Let me be very honest with you, Power BI wasn't at all on my to-do list to learn in 2020. I had learnt Microstrategy while I did my Masters in BI at ESC Clermont in France a few years back, and that seemed to work just fine for visualizing and playing with any datasets. But there was a sudden requirement in my job as a Data engineer to concentrate on the automatisation and industrialisation of Power BI reports, and I had to learn Power BI (fast!) in order to do my work well. Today, I am publishing dashboards and reports regularly for teams in 3 different continents, and it has been one rich and intense learning journey. In this article, I will try to go through some resources, sites, and tips that helped me to learn this tool quickly, and also some pitfalls and traps that could be avoided on this learning journey! ( I will assume here you have basic BI knowledge, you understand what is a data model for example, or what is a join between two tables, and that you know the difference between measures and columns. )

1) Get started with Guy in a Cube

One of the best places to get started is the Guy in a Cube channel on YouTube. Adam Saxton and Patrick create videos that are funny, interesting and target some very relevant issues in Power BI, from building a better data model, to calculating measures and the difference between filters and slicers and even on how to publish reports and share effectively. If you have never used Power BI before, I highly recommend this channel to get a feel of what the capabilities (and the limitations) of the tool are.

Then, start by looking at the series of mini-videos and modules on the Microsoft Learn page: https://powerbi.microsoft.com/en-us/learning/ . The videos are short and sweet, and the modules are very well structured and segmented. One of the most helpful ones I've found is how to publish and share reports , as you work between your local PC and the cloud service: https://docs.microsoft.com/fr-fr/learn/modules/publish-share-power-bi/. Using Power BI in an organisation entails having knowledge of the cloud service, Power BI apps and One Drive as well, so learning this part well helps a lot in the long run.

A super site for solving common Power BI problems : https://radacad.com/blog

Radacad, with regular content from two of Microsoft's top professionals in the fields of BI and AI, Reda Rad and Leila Etati, is by far one of the best and most helpful sites for solving common (and uncommon) data problems. Their blog posts cover many issues from data modélisation to DAX filtering, and is suitable for both beginners and experts alike. I went to their site to find innovative problems for solutions regarding performance improvement while using Calculate functions. However, you should already have some knowledge of working with Power BI ideally before visiting the site to get the most value.

2) Get inspired and practice on real datasets that are available online.

If you are not already using Power BI in your organisation, the best way is to train yourself on some dummy data available online. Power BI , being a Microsoft product, has loads of good documentation available online, and also sample reports which you can download. One of my favorites is the COVID 19 dataset: https://docs.microsoft.com/en-us/power-bi/create-reports/sample-covid-19-us, which is both a relevant topic and with data that is in a format that is easy to visualize and understand.

Power BI also has an excellent community page, which is good to both get inspired and to pose questions about doubts you might have about refresh, formulas, visualisations, and to find innovative solutions to a wide variety of problems. Have a look at: https://community.powerbi.com/

Finally, here are the 3 important lessons that I've learnt (sometimes the hard way) regarding actually building your Power BI report for good performance during the last year that I would like to share.

3) Start to draw your Data Model By HAND on a piece of paper. 

Age-old BI advice, but doesn't hurt to repeat it. Don't start off by creating your data model on the computer. When you draw your data model on paper, your brain (which is an amazing thing!) is able to come up with creative solutions to mapping between different tables, and allows to explore different possibilities for a good data model. Also, it is much better this way to explore the limitations of your data model, rather than creating it on the computer only to discard it later because you realized that there are some incompatibilities that you hadn't thought of.

4) Create a calendar table to manage your dates.

Dates can be a very tricky thing in Power BI, especially when you need to use dates for filtering, slicing and dicing your data in your report. A date table or calendar table with some parameters helps to have a centralized place where your dates are managed, which you can connect to your principal fact table in your data model in order to have effective filtering. It's quite simple to create a calendar table with just a few lines of DAX: imagine that if your main fact table is 30 million lines, but you have only 30 unique dates in this table. Better to have your filter query or visualisation look at 30 unique dates in a calendar table rather than scroll through 30 million lines in order to make a filter on a date.

5) Try to make all your data transformations, before having to resort to DAX if possible. 

Data Analysis Expression (DAX) language is Power BI's language to help make transformations, expressions and formulas, and is principally used for creating measures for dynamic calculation. For example, the sum of a sales quantity while applying some filters dynamically on a given period of time. (See: https://docs.microsoft.com/en-us/dax/dax-overview)

But try not to use DAX for creating columns for data transformation! This should be done instead in M-query on the Power Pivot side (using Transform Data/Edit Queries) or in Databricks if you are using that tool for data transformation.

Here is an excellent illustration of this principle, designed by Julien Bournat, my Advanced Data Modeling professor at ESC and my CGI manager. The idea is to do all transformations "en-amont" as we say in French , that is , before bringing the data into the actual Power BI desktop , and then using DAX just for calculating dynamic measures. This will bring great performance improvement if you stick to this principle !

6) Make a plan of learning for yourself

The best way to advance is to be systematic and make a learning schedule that is both realistic and feasible, with clearly defined goals. Power BI usually takes a month to understand the basics, another 2 months to understand the "edit queries and transformation editor" as well as learning basic DAX, and finally 3 months to really work on publishing quality reports. Afterwards, it's all about your journey to self-mastery.

And so, what are my personal plans for my Power BI learning in 2021 ? I need to complete an excellent Udemy course on Advanced DAX, get my professional certification in Power BI , and possibly teach some Power BI with CGI, if the opportunity presents itself. But more than anything, it's to continue to work with students, work colleagues and friends on solving Power BI problems, and this is what really permits me to grow my knowledge in this field.

Thanks for reading to the end Please like and comment if this article was inspiring for you, or you have other advice or your own Power BI learning journey to share.