VizWiz

Data Viz Done Right

December 3, 2019

To the Makeover Monday Community: Goodbye and Thank you!

No comments
Making over data visualizations has been my primary learning method for the past 10+ years. My first blog post was a simple makeover of a pie chart.

Before


After


It was nothing fancy (I didn't even know how to take a decent screenshot) and it was the start of something that has been a huge part of my life since. Fast forward to today (3 Dec 2019) and over 40% of my blog posts have been makeovers. 

My first “official” Makeover Monday was on 28 April 2014. 292 weeks have passed and I’ve completed 344 Makeover Mondays. I'm not boasts, rather I hope this shows you how with some focused time each week, your skills can improve very quickly.

My debut was a makeover about beer prices in Major League Baseball.

Before



After



Again, not great, but everyone starts somewhere. 

So why am I writing about this? Well, my time leading Makeover Monday has come to its natural end.

Leading Makeover Monday as a Community project for the last four years (2016 with Andy Cotgreave, 2017-2019 with Eva Murray) has been an incredibly rewarding experience for me. The fact that 1700 people turned up for a MM Live at TC19 simply astounds me. There have been thousands and thousands of vizzes created. Hundreds of people have used MM to get a new job. I was able to write a book with Eva; writing a book had been a goal of mine for a very long time.

I have lots of other things going on in my life that I want to give more attention to. For example,

  1. Spending time with my kids is my highest priority and even if it’s just a few more hours per week, that’s still a few more hours per week than I currently get with them. 
  2. I’m training for Challenge Roth, my first full distance triathlon. This will obviously take up A LOT of time. 
  3. I want more time to continue to create new content for The Data School. We have lots in store and I’m excited about the possibilities.

I’m going to miss you all, but I won’t be far away. I’m fairly certain I’ll participate every week. Feel free to tag me in your work.

Taking my place is the only person (other than me) that has completed EVERY Makeover Monday since 2016... Charlie Hutcheson. Give him a follow on Twitter and check out his incredible portfolio on Tableau Public (382 vizzes and counting). 

I believe this is the right time to give Charlie the exposure he deserves and let him shine. He’s a very close friend of mine and even lets me bully him into doing races with me that he clearly had no interest in doing! 

Charlie has been helping Eva with the Weekly Viz Reviews this year as often as he could. Charlie’s style, communication, and patience will continue to help Makeover Monday evolve.

I will miss you all and I’ll miss Makeover Monday. I have one lesson to leave you with,

FOCUS ON LEARNING

Learning more will never let you down.

Gratefully yours,
Andy

#TableauTipTuesday: How to show axis marks only at the top of a scatter plot

No comments
This week's tip is inspired by Workout Wednesday week 46 from the Tableau Conference. This challenge requires you to display the x-axis of the scatter plot on the top and not the bottom.

In this week's tip, I show you how to show the axis of a scatter plot only on the top of the view. Tableau doesn't provide an option to move the axis to the top, so this trick shows you a simple workaround.

Enjoy!

December 1, 2019

#MakeoverMonday: How have annual wages changed for union vs. non-union employees?

No comments
Four weeks to go with Makeover Monday 2019. We've had lots of interesting vizzes to makeover and lots of interesting data. This week, I wanted to pick a simple visualization and simple data.


What works well?

  • I like the handwriting font. It makes the viz look fun.
  • The colors are distinct enough.
  • Using shading on the title as a legend

What could be improved?

  • Some hands are holding another, some are not. What does that mean? Does two hands mean union? If so, I don't understand why they join where they do.
  • Using weekly wages is a tough concept to grasp. Why not convert it to annual wages?
  • The viz is clearly not designed for any sort of precision or comparison.

What I did

  • I really liked this Viz of the Day recently by Spencer Bauke and thought this was a good data set to try to emulate his work.
  • I wanted to use parameter actions to allow the user to change the comparison year.
  • I also wanted to use set actions like Spencer did, but this data wasn't structured in a way that made sense to try to do that.
  • This turned out to be very good practice for LOD expressions.
  • I loved using containers to lay all of this out!! It's a lot of work, but much easier to get everything to line up and all be the same size.

Here's my Makeover Monday week 49. Click on the image for the interactive version.

November 27, 2019

Five Essentials of Effective Metrics

No comments
I ran across this post I wrote on another blog where I used to write about project management. The original post is from February 2007, two months before I first downloaded Tableau.

Reading it again, the message about effective metrics still holds true today.  I've edited the post a bit to reflect data analysis projects rather than project management. At the time, I ran across a white paper that summarized the things we need to keep in mind for a metrics program. The paper didn't specify the metrics to collect, just the properties they should all have.

From my perspective, for any metric to be useful, it needs to help the stakeholder make decisions. All metrics should be actionable. If it's not actionable, then it's not useful.

Introduction

Information should be made available to all stakeholders throughout the lifecycle of a product. To be effective, metrics must be properly planned, managed and acted upon. What is measured, how it’s collected and how it’s interpreted are the difference between brilliant insights and blind alleys on the path to metrics-driven decision making.

The key is to ensure metrics are meaningful, up-to-date, unobtrusive, empirical and actionable.

#1 - MEANINGFUL

Metrics should focus on simple and fundamental units of measure for the given project. Understanding the key metrics across a portfolio of products can provide an important level of insight that enables organizations to understand opportunities and risks. It also provides a uniform basis for comparison across products, time, etc. Select metrics that will enable you to steer your company in a meaningful way.

#2 - UP-TO-DATE

It is important to look for metrics that can be captured automatically. Ensure that the metric is consistently based on up-to-date data.

#3 - UNOBTRUSIVE

The process of collecting data for your metrics program should be seamless and unobtrusive, not imposing new processes or asking stakeholders to spend time collecting or reporting on other data to get the answers to their questions.

#4 - EMPIRICAL

Metrics solutions should capture updated data as soon as reasonably possible, eliminating all of the issues that compromise the integrity and accuracy of data. Additionally, the use metrics that ensures data consistency; e.g., an working hour should be normalized to be the same in Boston, Bangalore, Mumbai and Beijing.

#5 - ACTIONABLE

It is critical that the metrics you gather inform specific decisions. Avoid information that is nice to know, but doesn’t help you make decisions or solve problems.

The litmus test for any metric is asking the question, “What decision or decisions does this metric inform?” Be sure you select your metrics based on a clear understanding of how actionable they are and be sure they are tied to a question you feel strongly you need to answer to effect the outcome.

It is also critically important to ensure that you are able to act on and react to metrics in a clear and meaningful way.

Finally, be sure that metrics are inclusive and that data is available to all stakeholders. Data that is widely available is empowering.

November 25, 2019

#MakeoverMonday: Where are the squirrels of Central Park?

No comments
When I saw the data set about New York's squirrels in the weekly email from Data is Plural, I knew it would be a fun data set for Makeover Monday and I had it all lined up for next week...that is until I went into our weekly planner and saw that Eva had already chosen it for this week.

Let's have a look at the original viz:


What works well?

  • A map is a good way to represent the location of the squirrels.
  • Showing the details on the map, like the ponds in the park and the roads around the park, help provide context.
  • The dots are easy to see against the background.
  • The title and description help explain the data.

What could be improved?

  • There are some red dots. What do those mean?
  • It would be interesting to see when the squirrels are spotted. Are there more in the winter? How do their habits change?

And here's my makeover. Enjoy!

November 19, 2019

#TableauTipTuesday: How to Create a Pie Chart Drilldown

No comments
This isn't something I can condone the use of, but I thought it was a fun technique for learning set actions. In this example, I show you how to create a pie chart inside of a pie chart where the inner pie chart will drill down based on the selection of the outer pie chart.

November 18, 2019

#MakeoverMonday: Tween and Teen Smartphone Ownership

No comments
This week, we're looking at the change in smartphone ownership for tweens and teens between 2015 and 2019.

SOURCE: Morning Brew

What works well?

  • Clear title
  • X-axis is clearly labeled
  • Including the data source
  • Colors are easy to distinguish
  • Vertical lines help draw the eye to compare the years within each age
  • Including labels since the y-axis is hidden

What could be improved?

  • The title could be less bold.
  • The title uses the color for 2015, but it's not related to one year only.
  • The dots are distracting since they are so large.
  • The labels are helpful, but do they need to be so big?
  • With the vertical lines connecting the dots within the year, and the line connecting the ages across the years, I'm not sure which is more important. Given the title, the focus seems like it should be on comparing years within an age.
  • The vertical lines don't need to be so broad.

What I did

  • Removed the lines to make the focus comparing the ownership within an age group
  • Surrounded the dots with a band to ensure the user reads the data within each age group
  • Colored the bands by the change to accentuate the ages that have changed the most
  • Included the labels, but made them very small as to not distract from the analysis
  • Created a mobile version for practice

November 12, 2019

#MakeoverMonday: Literacy Rates Around the World

No comments
It's TC19 week and Eva has provided a data set about literacy rates around the World for the 1200+ people in Vegas to viz live plus the hundreds more of you that aren't live with us.

Here's the original viz:


What works well?

  • The data by region is ordered alphabetically, making it easy to find each region.
  • The bar chart is sorted by largest to smallest.
  • Nice filtering options

What could be improved?

  • A diverging color palette should only be used when there is a logical midpoint or goal. I don't see those in this viz.
  • The squares are hard to understand.
  • I don't find the map very useful. It would be more useful if it zoomed in when a region is selected.
  • There's no title.
  • There's too much text.
  • The bar chart seems to go out past the edge, or at least visually it appears that way.

What I did

  • I created a KPI scorecard so that I could understand the patterns for the overall or an individual country. Are literacy rates improving or regressing?
  • Show the distribution of the rates of the countries within each region
  • Within each region, which countries are above or below the median for that region?
  • How has the literacy rate changed over time?
  • Allow simple filtering options.

I drew inspiration from Workout Wednesday week 51 2018: Container Fun from Rody Zakovich. I love finding reasons to practice techniques I've tried before and want to master. Consider challenging yourself to learn something new each week.

Enjoy!

November 5, 2019

#TableauTipTuesday: Using Level of Detail Expressions to Count Items Exceeding a Threshold

No comments
In this tip, I show three examples of using Level of Detail expressions to count members of a dimension in a view. I also show how parameters can be used for counting members based on thresholds.

I ended up babbling quite a bit as I created more examples; sorry for that, but I was on a roll.

November 4, 2019

How Many Rats Are Near Hungry Cat?

No comments
Well, I don't really mean near YOU, I mean near people that live in New York. One of the fun datasets we play with at The Data School when I'm teaching them spatial analytics is Rat Sightings in New York.

And right after I taught this class to DS16, Lorna Eden posted the Workout Wednesday week 43 challenge. In this challenge, you had to find all casinos within X miles of a casino you click on. This required using the new DISTANCE function that came into Tableau 2019.3.1.

So, why not practice this technique more, but with rats? Instead of clicking on a casino, you can click on a rat to make it the Hungry Cat and find all rats within X miles of the cat. Silly, yes, and fun to practice too. The rats all have names too.

Lastly, I wanted to resize the dots based on the number of rats in the view. I used this blog post from The Data School, except I used an LOD instead of a table calc.

Enjoy! Find the rats near you.

November 3, 2019

#MakeoverMonday: Is Las Vegas Convention Attendance a Recession Indicator?

No comments
Tableau Conference is headed to Vegas shortly, so I thought for Makeover Monday week 45 we should look at data about visitors to Las Vegas. The original viz comes from Calculated Risk:

SOURCE: Calculated Risk

What works well?
  • The time goes from oldest to latest.
  • The colors are easy to distinguish.
  • The axes are well labeled.
  • Including the caveat for 2019 since it's not a complete year in the data.

What could be improved?
  • The axes aren't synchronized; I'd like to see how they would look synchronized.
  • Without referring back to the color legend, I don't know which axis goes with which metric.
  • Using a dual axis chart implies there's a correlation between the two measures. There might be, but it could be displayed other way to make that more evident.
  • There no indicator of the data source.

What I did
I started by reading the original blog post. What caught my attention in particular was the last sentence:
Historically, declines in Las Vegas visitor traffic have been associated with economic weakness, so the slight declines over the last two years was concerning.

Super interesting! So this is where my worked started. I first annualized the data to make 2019 comparable to the rest of the years. From there, I created a connected scatterplot, which takes the two metrics in the original chart, plots one on the x-axis and the other on the y-axis, and connect the points by the year. This lead to a swirly look at the end, which made the relationship difficult to understand.

Instead, I chose to focus on the "red" line of the original, i.e., convention visitors. I wanted to see if convention visitors was indeed a recession indicator. The chart was simple to make, then some googling turned up the recession dates. Low and behold, convention visitors to Vegas sure do look like a leading indicator for a recession. If this is true, then we're on the verge of a recession very soon.

Click on the image for the interactive version.

October 22, 2019

#TableauTipTuesday: Using Distribution Lines to Provide Space for Labels

No comments
Credit for this tip goes to Rody Zakovich (@RodyZakovich). In the past, I've always created complicated table calcs to give labels room above the max and below the min of a line chart.

With distribution lines, you no longer need to do that. Simply set a percentage offset and you're good to go! So simple!

October 20, 2019

#MakeoverMonday: The Age at Which Most People Are Dying by Suicide Has Increased Over Time

No comments
We're heading into the home stretch for #MakeoverMonday 2019. This week, I chose a visualization from the ONS about the age at which people commit suicide in the UK. I picked this data set for two primary reason:

  1. It's a simple data set.
  2. The original visualization is really good.

Let's have a look.


What Works Well?
  • The patterns are super intuitive to see and understand.
  • Amazing interactivity that allows you to look at an entire year without the overlap of the other years.
  • Nice annotations
  • Years and ages are in the appropriate order
  • Using the intensity of a single color to emphasize the number of suicides

What could be improved?
  • The number of axis ticks could be reduced to every five or ten years.

What I did
I thought I would try to build a horizon chart, but it became too complex and the data wasn't quite suited for it. I then built a series of area charts like the original, but without the overlap. It didn't look good; the original has nice rounded lines, whereas a Tableau area chart has sharp edges.

I tried line charts, small multiples, running totals, none of them worked. I ended up keeping it very simple and went with a heatmap that I think gets pretty close to the same analysis as the original.

October 14, 2019

#MakeoverMonday: Ironman World Championship Medalists

No comments
Sunday was the 43rd Iron World Championship. It was the first time I spent a lot of time watching it and I found it pretty cool. I figured I should pay more attention to how it all works given I'm doing Challenge Roth in July. I'm a bit daunted by the prospect of competing in an Ironman distance, but I know I can do it with proper training.

The viz to makeover is a simple table from Wikipedia:


What works well?
  • The years are listed in order from most recent to oldest.
  • Table can be sorted
  • Including separate columns for each medal
  • Including links to each athlete

What could be improved?
  • I'm not convinced that both the flag and country abbreviations are necessary; one is probably enough.
  • Some of the athletes have red text and some have blue. I couldn't find anything on the page that explained this.
  • Comparing athletes across years is difficult because of the precision of the times/

What I did
After exploring the data for a few minutes, I remembered that Rody Zakovich created an incredible viz about the Winter Olympics (check it out here) and I've been wanting to emulate it. This data set proved perfect for it. This is the beauty of Tableau Public; you can download workbooks, see how someone created their work, and use it to help create your own.

Here's my viz for Makeover Monday week 42 (click on the image for the interactive version).

October 8, 2019

#TableauTipTuesday: How to Automatically Apply Worksheet Actions as Dashboard Actions

No comments
This week's tip was brought to my attention by Zen Master Rosario Gauna. One of the really annoying things about Tableau for the longest time has been needing to create a separate dashboard action that mimics a worksheet action. In other words, you create a worksheet action, you like it, then when you add the sheet to a dashboard, the action is no longer there and you have to create it again.

Well in this tip, I show you how to make a worksheet action automatically apply to a dashboard action without having to create a separate dashboard action. Genius!

October 7, 2019

#MakeoverMonday: Bearwood Corporate Services - The Money Behind David Cameron's Conservative Party

No comments
What a fascinating data set! More on that in a minute. First, let's have a look at this week's viz to makeover. It comes from The Electoral Commission and focuses on political donations in the UK, which are required to be reported.

SOURCE: THE ELECTORAL COMMISSION

WHAT WORKS WELL?
  • Placing the filters on the upper right let me know immediately that I can interact with the data to find my own story.
  • The bar chart is sorted in descending order.
  • The summary numbers provide some context, but not much.

WHAT COULD BE IMPROVED?
  • The bar chart would be easier to read if it was horizontal.
  • Why are all of the bars colored? There are way too many colors and they have no meaning.
  • The packed bubbles would be much better as a bar chart or BANs.

WHAT I DID
I started by exploring each field in the data set. Many of them didn't have information I found useful, so I hid all of those fields so that they would not distract from my analysis. 

As I explored the donors and who gave what to whom, I saw that Bearwood Corporate Services was donating A LOT to the Conservative Party over a period of a few years. I hadn't heard of it before so I did some research on Google and it turns out that they were pretty controversial and closely linked to the rise to power of David Cameron.

That's where my story begins...

September 30, 2019

#MakeoverMonday: London's Aging Population

No comments
This week, Eva chose a viz/data from the London Datastore about population projections. Here's the original viz:

SOURCE: London Datastore

What works well?
  • It's a simple bar chart, which is very easy to read.
  • Uses a single color; we often see the bars double encoded with the same value as the length of the bar.
  • The axis starts at zero.
  • Simple, clear axis labels

What could be improved?
  • I think a line chart would be even easier to read.
  • The title says "Population of London" but it's really the projected population plus the past population; some clarification would be good.
  • The legend isn't needed.
  • The y-axis title could either be removed or changed. "Number" doesn't mean a whole lot.

For my makeover, I was interested in comparing the distribution of the population by age for one year compared to the distribution of the population in 2050. For example, what was % of the population for 45 year olds for 2018 and 2050, then compare those two values. This then shows how the population distribution will change.