How To: Shaded Slope Charts in Tableau

I recently found an article showcasing 2016 in charts, and I really liked the slope charts with the area shaded between the two categories being compared.

I found this to be a really cool technique for making a point and telling the story.  I wanted to see if I could recreate this in Tableau, so here's how it's done.  I'm using the Superstore data in this example.

1.  Create a standard slope chart by setting the Year to discrete, filtering to two discrete years, adding the measure to the Rows and Category to the detail.

2. In this case, I want to show the difference in Profit between Technology and Furniture.  To create the shaded area, I need to isolate their respective profits.

3. Drag the newly-created Furniture Profit measure to the Rows, and then drag the Technology Profit pill to the left to create the Measure Values pill.

4. Create a dual axis and synchronize the axes, change the Measure Values to Area, and remove Category from the detail.

5. Move the Technology Profit measure to the top of the Measure Values list and change the Furniture Profit color to white.  I also went ahead and changed the Technology Profit color to blue.

6. Since we are using an area chart, the Technology Profit is being stacked on top of the Furniture Profit, which is pushing the shaded area up above the slope chart even when synchronized.  To fix this, we need to subtract the Furniture Profit from the Technology Profit.  I went ahead and edited the calculation in the shelf: SUM([Technology Profit]])-SUM([Furniture Profit]])

7. To color the dimensions I wanted to compare, I created a calculation to only color those dimensions.  Drag this dimension to the color and turn on the dotted line.

8. Hide headers and clean up the chart formatting. I also thickened the lines and set the area chart color to blue with 20% transparency in my example.

9. Finally, incorporate storytelling elements like using color to point out the categories being compared and annotations to highlight the gap.

Feel free to download the workbook and comment with any questions.

National Championship 2017: Clemson vs Alabama

As a kid, I vividly remember seeing a Clemson 1981 National Championship poster in my friend's room.  I remember seeing that poster and being blow away that Clemson had previously won a national championship.  I've pondered over time if Clemson would ever win a National Championship in my lifetime; 1981 was a few years before I was born.  I always said that if somehow Clemson ever made it to the National Championship I would attend the game, and I was lucky enough to go to both of them and ultimately see the big win this year.  As an alum and employee of Clemson University, Clemson has always been a huge part of my life, so that whole experience is something I will remember forever.

To commemorate this event, I wanted to create a visualization of the entire game that would work well as a poster.  I plan on having this printed out for my son's room at some point.  I wanted to make it interactive so that you can hover over any play and see the details.  I also wanted to create something that would allow you to see the actual play in the game, so I was able to link game footage to all of the data in the visualization.

Hover over any play to see details; Click on any play to see the actual play in the game.

Click for the full interactive version

Makeover Monday: Australia's Gender Pay Gap

For Makeover Monday, I wanted to take the data referenced in this article about Australia's wage gap and create a visualization that would quickly show the gap between wages for men and women in Australia.

To begin with, I wanted to create a visual that would emphasize the gaps in pay by occupation.  Previously, I've used DNA or barbell charts, but this time I wanted to try out using Gantt bars to show variance.

When building the visualization, I did have some questions about using the mean instead of the median.  Since we can't see the underlying data, one has to wonder if there are any outliers that would skew the data.  Would it be more appropriate to use a median here, or is the mean okay?

This issue has been raised before about visualizing data you didn't create or curate, so I thought I'd take a quick glance.  If you look at the source data, it comes from Australian tax data, which is compiled from tax returns.  This is taken from the entire population of Australia, so it is not a sample.  Since it is an entire population, the average is generally acceptable to use, so that's why I went ahead and used it.  Also, the Australian government does not report the median in their statistics either.

Path to the Playoff: 2016 College Football Weekly Rankings

I've created this type of visualization for previous years AP Rankings, but I wanted to do an updated version for the College Football playoff ranking.  I'd recently seen a post by Rody Zakovich on using sigmoids in bump charts, and I wanted to see if I could apply it to my design.

With this viz, I wanted to highlight the path of the four teams that have made the playoff.

I also wanted to give users the ability to highlight their favorite team.

Additionally, I wanted to add as much detail as possible to the tooltips.  Notice the indicator for teams rising or falling in the ranking.

I found it interesting to the see the path of the teams that missed out on the playoffs.  There was a lot of controversy with Penn State not getting in while Ohio State did.  It was interesting to see teams like Penn State and Oklahoma State make a run to the top 4 after being outside of the top 10 in the initial ranking.  You have to wonder if it is possible to overcome an initial low ranking in the poll, but I also expect to see the playoff expand to 8 teams over time.