Let me start this article with a definitive statement: I am definitely qualified to write this article about Power BI. My qualifications do not have anything to do with that software product, however. My work history and knowledge of Microsoft products are the reason I feel that I am qualified to write this piece.
There is no question that my 30 years of experience using Microsoft products puts me in the category of someone who knows what they are talking about. I might even be considered a “highly-trained monkey worker” when it comes to using Microsoft products. If someone were to ask me what is my most refined analytics skill set, I’d say it is working in Excel and writing custom VBA codes. Those are a couple of reasons why I feel that I am qualified to offer an opinion on Power BI.
I have done more scientific, analytical, and numerical work using Microsoft products than I have done in all other computer-based tools I have ever used, even when all the other tools are combined. I have created hundreds (quite possibly thousands) of Excel worksheets, graphics, and many multi-thousand line VBA codes. I have programmed in too many computer languages for me to accurately remember, and a lot of it was done using Microsoft compilers.
For these reasons, what I am about to say should be considered an opinion that has some merit. I would also like to say that I have a huge amount of respect for Microsoft, so the statements I make in this article were very carefully considered before I wrote them and they are not intended to be mean-spirited. I hereby reserve the right to have my opinions altered.
I have always written articles in this blog with candor, honesty, passion, and with a vision of the way I see things. I am not afraid of telling the truth, even when the truth hurts.
My Motivations For This Article
I really only have two motivations for this article. First, I like to document my thoughts at various points in time so that I can look back and see where I was in my understanding over time. I like to be able to look in the rear-view mirror to see where I was, knowing where I am now.
I don’t have any connection to any computer software firms, so my statements are simply my opinions that are based on a continuous work history in both science and business over a long time period. This article was written for me, and only me. I just decided to share it with you.
Secondly, I like to be a person that continuously improves. I like the challenge of learning new tools and concepts and I like sharing my insights. I also like to help other people learn by writing these articles. People who know me have come to realize that I am not afraid to try new things.
Finally, I know that my opinions in this article may not be popular with a lot of people. There are many wonderful, highly intelligent people working for Microsoft, consulting firms and companies that are using Power BI to do great things. To these people, I applaud your efforts. However, for me to begin using Power BI would require that Power BI be a much better tool than the tools I already use. If you want to find out the answer to whether this is the case, you will need to read this article.
Going Back in Time
It was 1993 when the crazy movie called Groundhog Day was released. In the movie, Bill Murry finds himself stuck in a time loop, where he relives the same day over and over again. Needless to say, he quickly became frustrated with his situation because he feels stuck in time. Today, I felt like Bill Murray’s character.
Today, I found myself experiencing various forms of Groundhog Day as I tried to learn how to use Power BI. I will explain exactly what I mean so that you can understand my points of view. I will say that this experience has proven to me that “life has a funny way of repeating itself”.
Day 1 of the Training Class
Today I spent a week in a Power BI training class that was planned too ambitiously. The goal of the class was to nearly cover the entire contents of a 140 page training document. This document took us through the development of a DIAD, or a Dashboard in a Day. The example that is used appears to be a standard set of files developed to train new users of Power BI.
The two-day training course is designed to complete the DIAD as well as to cover various other topics of interest, such as the Azure Machine Learning part of Power BI. This part of the software can be used to determine Quick Insights hidden within your data.
The Quick Insights feature is built on a growing set of advanced analytical algorithms developed in conjunction with Microsoft Research that are designed to help people to find insights in their data in new ways. I found this to be a great idea and hoped that it would be a new technique for me to expand my skill set.
The Quick Insights advanced algorithms include the following approaches:
- Majority (Major factors)
- Category outliers (top/bottom)
- Time series outliers
- Overall trends in time series
- Seasonality in time series
- Steady share
- Change points in a time series
I was particularly interested in testing this software feature because I pursued a lot of these topics when I was working as a process improvement consultant. To use Quick Insights, you have to use the cloud-based Power BI package. I did this today and I’ll explain my findings later in this article.
Now as I said yesterday, let the games begin.
The Fundamental Flaws In The Training Course
I have been teaching quantitative college courses for a long time, and I have been teaching people to use Tableau for at least the past 5 years. I have a lot of experience directly related to teaching computer-based courses, and for today I was a student in one of those courses.
First of all, when you teach a new software product to people, there are two things you must do. If these two things are not done properly, people will not learn the software product in an optimal way and they will leave frustrated and unsure of what they were trying to learn. The teacher must, at a minimum, do the following:
- Explain the fundamentals of the software package to help people understand how it works
- Start with simple concepts and build towards complexity throughout the course.
Neither of those things happened today. There were were no discussions of fundamentals and the example problem used is simply too complex for most beginners. In fact, the structural design of the course is so fundamentally flawed that I have decided to not discuss this topic further.
Once I determined that the course was not going to help me learn Power BI (after about 3 agonizing hours), I embarked on my own work. I made some good progress that helped me formulate the opinions in this article.
However, before leaving this topic, I will offer this insight.
In February 2008, when I first used Tableau, I realized nearly instantly that Tableau was a paradigm-busting approach to visual analytics. For more information on why I believe this, please read this article that I wrote 2 years ago.
I knew how special Tableau was because I used to write my own graphical processing programs for numerical models. If you have ever tried to write these types of programs, you know how hard it is to do correctly.
The reason that Tableau blew me out of the water was this simple fact: I spent almost no time setting-up and formatting graphics and almost all my time was spent on analysis. Furthermore, the graphics and dashboards were always natively beautiful.
Groundhog Day #1
Today, was Groundhog day #1 for the following reasons. Whereas Tableau has provided to me the ability to rapidly analyze data, I spent all day today setting-up and formatting visuals in the Power BI example.
There was no time for data analysis during the course because maximum effort was needed to simply create the graphics. Since I have spent over 30 years doing scientific and business studies that require data visualization, I was not satisfied with Power BI’s ability to perform rapid visual analysis!
The groundhog day occurred because I was experiencing deja vu for most of the morning. I found myself executing countless graphics configuration operations. I felt like the software required me to work like I did a long time ago. I found myself recreating the steps needed to build Excel graphics from the 1990s. In fact, I overheard one of other students make this comment:
This would have been cool back in 2005.
I think my feeling of deja vu was most strongly triggered by how Power BI forces the user to do so many operations to accomplish the most simple things. It is very apparent to me that much of the Excel graphical framework, with countless configuration settings, has been carried forward into Power BI. On one hand it is great to have this level of control. On the other hand, it sucks to have to do so many things to see your data.
I’m not saying that Power BI is as arduous as creating charts in Excel was for me 20 or even 10 years ago, but the conceptual software design and primitive thinking has definitely been carried forward.
I believe that Microsoft has missed a great opportunity to change the paradigm of their visual platform to something that would be considered consistent with contemporary professional standards. I think they did this because they were so far behind the leaders in the field that they rushed their product development simply to get “in the game” before it was too late to compete.
Groundhog Day #2
I got stuck in another time loop today when our example dashboard required me to configure conditional formatting. I thought to myself: Are we really still using conditional formatting with all of its rules and ugly colors? I knew that part of the course was bad when I heard a student say:
They are telling us to use green as a color, but I can’t find green – I only see blue!
I think I have gotten very spoiled by the ease of use of Tableau. When I thought about configuring colors in such a way, I think my brain rebelled. This approach seems like such an archaic concept that I couldn’t process the instructions and got stuck in an infinite time loop.
Groundhog Day #3
As if GHD #1 and #2 weren’t bad enough, I knew I was in trouble when the following occurred.
We loaded a bunch of data and I had no idea how the data sets would be connected. After we loaded the data, we had to rename fields, change data types for no apparent reason (like changing a floating point number to a fixed digit floating point number), and do a bunch of other operations before we could even look at the data. The next step was to pick the type of graphic we wanted.
Well, the problem was, the required data connections were not in place for us to be able to visualize that type of chart! We had to stop, turn around, and go back to the starting line in the data model to make the appropriate data connections before we were allowed to move forward once again and see the data. I felt like I was on a data merry-go-round.
I stopped having to do this type of work a long time ago. Even with the instructions provided, the clarity of what we were actually doing was about that of a Jackson Pollack painting. I knew it was bad when I heard another student say this:
They can’t really ask us to use this instead of Tableau, can they? No, there is no way they are going to ask us to do that!
Groundhog Day #4
After GHD #3, I had an epiphany. I looked around the room and on the video conference monitors, I saw a bunch of furled brows, hands on foreheads, and basically disengaged people being trained. After spending hours wrangling through this tangled web of instructions, missteps, and unknown software errors, I suddenly had the following thought:
If it takes a 140 pages of instructions to build a dashboard in a day, there is something seriously wrong.
I could have done all this work in Alteryx and Tableau in about 1/2 an hour, with more beauty, insight and way more awesome dashboards than anything that can possibly be created in Power BI. I thought to myself, if I have to do this ever again, I might have to pull a Bill Murray and take myself out of the game because tomorrow will come and I can start again with Alteryx and Tableau.
I just can’t imagine using Power BI now, next year, or anytime soon thereafter. I refuse to go backwards in technology because I want to push the frontier of what is possible.
The relatively simplistic quality of the Power BI graphics created, the chaotic dashboards, the haphazard work methods and the incessant trivialities make working with this product about as much fun as having a molar removed. In fact, this course today made me feel like I did when the following video was taken.
In my opinion, today wasn’t a good day for Power BI. Yesterday, I was delighted in my first 5 minutes of usage. Today, the 8 hours I spent getting to know Power BI were not great. Tomorrow is another day, however, and I assure you that I still have an open mind that is willing to be convinced that this is a tool for me to use!
By the time I finished writing this article, I didn’t have the heart to share with you my “Quick Insights” that were “discovered” from one of my favorite data sets – the 3danim8 blogging experiment. I just figured it wouldn’t be fair for me to add insult to injury.
You never know what could happen tomorrow as the game can swing in the other direction with the right set of circumstances occurring. I think Power BI is here to stay because of Microsoft’s deep pockets and ambitions. Surely, it will get better, but for me, I don’t have years to wait for this to happen. I need to perform now, and Alteryx and Tableau allow me to do that better than anything else that exists.
Stay tuned for tomorrow’s episode, which likely won’t be published for a few days. A man has to work, after all! Thanks for reading.