Hands-on Data Visualization practice week 1

Taking part in #MakeoverMonday through Design Thinking

Every week, I’ll be taking part in #MakeoverMonday in which I’ll revise a bad visualization to practice creating better data visualizations. I’ll walk you through my process as a designer in revising my visualizations, including why I did things and my thought process in creating better visualizations based on Design Thinking.

Thie first practice will be free, as will one data visualization practice a month. Additional weeks will only be available to paid subscribers.

Visualization to revise

Walking you through my Design process

The first thing that I do, when looking at the chart is to understand what does and doesn’t work.

What does and doesn't work

The first thing that stood out to me was that this was a comparison between two values, the Bike values of 2019 and 2020, but the visualization wasn't a good one for comparison (line chart). It was clear that there were more bikes counted in 2020, but drawing an individual comparison of a single point might be incredibly hard.

In addition this dataset involved time periods before and after Covid, yet didn't seek to highlight it any way. Therefore, I wanted to highlight the impact that Covid had by showcasing how the # of bikes counted changed pre and post Covid.

Therefore, I went back to the dataset and worked on trying to come up with a better way of visualizaing the message.

Understanding the story I want to tell with the data

Based on what I thought the chart was telling me, I was telling the following data story:

“There was a change after Covid to the # of bikes counted on trails. To show this, we should compare data from pre-Covid declaration (2019) to Covid period (2020)”.

Therefore, I need to use a chart that’s easier to compare between the 2019 and 2020 values: a bar chart.

Given the type of story that I’m telling, I also need to revise the axes. I’m okay with having the week # as part of the X-axis as the dataset is tracking these numbers week by week, but I don’t think having the y-axis being total # of bikes is that useful.

Part of this is how messy the scale is, but the other part is that I’m more concerned about the reader understanding the magnitude of change between 2019 and 2020 rather then the number of bikes in each year. As a result, I need a better way of understanding the difference between the two.

How I made it better

I created my revised visualizaton in Tableau to gain more practice with the software. You can see a couple of changes with the chart.

When I was exploring the provided dataset, I found a value (change from 2019 to 2020) that was useful for my story: this value was a percentage increase (or decrease) based on the number of bikes in each week, comparing 2019 to 2020.

A number of different things changed as a result. The first is that Y-Axis, which had once been crowded with numbers, changed into an evenly spaced Percentage marker. Because percentages aren’t what comes to mind when thinking about scales of measurement, I also took it upon myself to label one of the largest data points, the first week of Covid restrictions, with how big the percentage increase was.

To make the bike counts pre and post Covid easier to compare, I highlighted pre and post Covid declaration in two colors, Blue and Orange, to showcase the effect on biking that Covid lockdowns had.

How I might improve this further

While the visualization was improved a fair amount, there are a number of things that I could do to further improve this if I was allotted more time and resources.

One of the first things I would do would be to bring another dataset, specifically about Remote Work in the UK, into the equation. It might add an additional layer of depth as it comes to understanding why there was such a huge boom on one particular week.

The other thing I might want to do is incorporate other aspects of the dataset if I decided to shift my focus. In addition to biking, the dataset also had the pedestrian counts for 2020 (i.e. hikers), as well as the change from 2019 to 2020. This might be a useful thing to look at given more time.