Dr.CurveNoMore or: How I learned to stop curving lines and love the bar chart.

For anyone that has followed me over the past year, I am about to contradict everything I use to do and say......

One of the things I love the most about Data Visualization is the fact that I don't know everything, and probably never will, which means there's always opportunity to learn. But, sometimes I have to learn things the hard way. For example, learning that making curvy lines or flashy chart types for the sake of doing them is not good practice, sometimes a simple bar chart can tell the story (and be beautiful too).

When I started my career as a Data Analyst, most of my visualizations were very basic. I used plain text, bar/lines charts and text tables. Why? Because at that point, that's really all I knew how to do. I doubt anyone starting with Tableau, or any other tool, jumps right into making Sankey diagrams. At the time, I thought what I was doing was really good too. I was taking data, and showing it visually in way that people could easily comprehend. Now, these early visualizations were not amazing, and the designs definitely needed improvement, but my main goal was always to display the data in the most meaningful way I could. Below is an example of my first Makeover Monday Viz.

This is not an amazing viz, it isn't very engaging, and it's a little repetitive, but it's pretty clear to read. I didn't use any Curves, or images, or crazy techniques, I just took the data and tried to visualize it in an easy to read form given the skills I had at the time.

I continued to make visualizations like this for while. And I felt I was getting better at making my designs be more professional, while also making them clean and engaging. Here is another one I did later into the year on Global Warming.

Once again, not a complex design. It has issues, but my objective was to clearly show the data, and I think I did a fair job at that. So my early work was very basic, I was never trying to "Wow", I was just doing the best I could at the time.

Then.....one fateful day, I was looking for data and created a dataset on all the words said in South Park Season 1. I had a lot of fun with the data, and decided that I wanted to try something different since the data didn't have any meaning. So when looking at the data, I decided to focus on the Words said by the Boys, and what percentage of those were Swear Words (According to a Swear word Database I found). In doing this, I decided I wanted to show how the Swear Words were distributed among the Boys, and I thought a Sankey Diagram would be a fun way to show it. So I found Chris Love's tutorial, and made the viz.

I was really happy with how this one turned out. I thought I used colors well, it was informative, and I implemented a cool new char, that I thought worked well for the story. After I made this, I posted it on Reddit, and then something happened that would change my personal vizzing style for a while. The viz went viral, it got over 250K hits in two days. I was ecstatic! Then my brain did something that was not too great. See, this wasn't the first viz I posted on Reddit, in fact I posted a lot of vizzes on Reddit, but none of them got more than a few thousand hits. So when this went Viral, I immediately said to myself, "Well, I never used a Sankey before, I used one here, and the viz went viral.....therefore, Sankeys must be a good chart". 

My brain forgot the first rule of Data Analysis, Correlation does not equal Causation. But I was too excited about the attention I was getting to realize it. Looking back at it now, I see most of the comments and hits were related to the Bar Charts and Area Charts, not the Sankey. People were just generally interested who said Curse words the most, and what percentage of words were Swears. But I didn't see that then, I just saw what I had did different from previous visualizations. 

Before long, I found myself trying to incorporate curves and unique chart types into every viz I did. Thing was, I wasn't doing it because I thought it was the best way to display the data, I was doing it to get views, to get attention. Worst part..... it was working. I started getting a lot of attention for my vizzes, and was getting recognized by the community. 

Now, when I started working at as a Data Analyst, one of the first things I did was read Information Dashboard Design by Stephen Few. The principles I learned from that book were always in my head. I always had the notion of "Best Practices". So when I was questioned about my "Curvy" vizzes, I started to make excuses, "Oh I do exec dashboards all day at work, this is fun time", or "I'm just practicing Tableau". But, as I've come to realize, I was just lying to myself. I wasn't using curves or images because I thought it was a good way to visualize the data, I was using it to get attention.

I continued on this path for quite some time. And I actually made some visualizations that were actually both engaging and informative, but as time continued, I just started to make more and more charts with the notion, "How do I make this viz Pop, how do I make people say Wow". I forgot how to be analyst. So what do I mean by this?

This visualization I did on Queen looks pretty cool, lot of unique and artistic design went into it. But can anyone actually learn anything from it? Seriously, can you look at this chart as an image, and pull any meaning out of it? Probably not, because I didn't care to make it easy to read, I just wanted to make it look cool. Like I said before, sometimes you have learn things the hard way.

One might be asking how I came about this revelation. How did I break myself of this bad habit. Couple of things.

First, Chris Love. 

Chris was one of the people who always critiqued my work, and challenged me on why I was doing curves. There were times I was a little angry by this, but only because subconsciously, I knew he was right. Looking back on it, I realize he was pushing me because he was trying to help me. Trying to help me realize that I didn't have to always make flashy vizzes, I could just Keep It Simple and tell the story, like I use to. I really appreciate everything he did, even if I didn't at the time he was doing it.

Second, Adam Crahen.

I had the privilege of working with Adam Crahen over at DOI. And, while my time working with him was short, I learned a lot! Adam is one hell of a Data Analyst, and he can make some stellar visualizations. Working with himmade me realize that Big Text and Bar charts can look pretty awesome, if you just clean it up and structure it the right way. You can make things simple, and engaging at the same time. 

Finally, I took a break.

I played Golf in High School and College, it is probably the only hobby I love more than vizzing. I use to practice and play hour and hours every day. Sometimes though, I wound get into a funk, I'd pick up a bad habit, and no matter how hard I worked, I couldn't kick it. The only thing that actually worked to kick the habit, was taking a break. When I first began to notice what I was doing, I quickly realized the only way I was going to kick the habit was to stop, so I did. I stopped vizzing and being really active in the community, and decided to do some reading. Broke out all the Few, Tufte, Cairo, Knaffic and Kirk books I had and started studying. If you ever find yourself in a funk, I highly suggest doing this, it works wonders!

So after all that, I finally had my revelation. I learned the error of my ways, and am here to say openly, I was wrong. Also, I am not implying that curves are bad, and I'm not saying anyone who uses them are doing it for attention, there are people who have mastered the technique of artistic and analytical design. But I am saying the reason I was using them were wrong. I may use curves again, but when I do, they will be for the right reasons.

Where am I at now?

I'm getting back to the basics. making the first 95% of my time dedicated to creating an effective visualization, and then spending the final 5% making it look good. So I'm back to square one, using Bar/Line Charts, Big Text and Text Tables. But now I understand the importance of why I'm doing it. You don't have to wow people, you don't have to use crazy techniques to make a viz engaging. Sometimes you just need to show the data in a clear way. Basic charts, like the Bar, are great charts. They are easy to read, they are flexible and you can make them look good. So here is a recent Makeover Monday submission. It isn't as fancy as the Queen viz, but it is a better viz.

I still have a lot to learn, and can guarantee that I will make mistakes in the future, but I've learned an important lesson. And I've learned to stop curving lines to get attention, and Love the Bar Chart.

-Rody Zakovich