The Friday project – week 9

Simplicity v Complexity

This week I created two vizzes, and I am not quite sure I quite nailed either one. With the benefit of hindsight, one was too simple, and one was too complex.

Makeover Monday

The first was my redesign for Makeover Monday to show Andy Kriebel’s Amex spend in 2016. I went for a simple, stripped-back approach, showing cumulative spend over the year as an area graph.

How much did Andy spend on his Amex in 2016?

 

amex-spend

So why do I think it is too simple? Because there are a number of questions which could be answered from the data which aren’t addressed, e.g.

  • how many transactions did Andy make?
  • what type of spend was it?
  • what days of the week did Andy spend the most?

If I get the time I will look to expand the viz to incorporate answers to these questions: to keep the design simple, but not the message.

Sankey experiment

The other viz I created this week was this Sankey diagram, using guidance from Chris Love.

Sankey diagram (technical experiment)

 

sankey

So why do I think it is too complex?

Firstly the graph type is not a standard one in Tableau, and the method Chris has used is very complex – I would never have worked out how to do it on my own.

Secondly for this type of Sankey, where there are distinct categories on the left and the right, there are better, simpler ways of showing the breakdown in volume, as Chris himself explains in his excellent talk on embracing simplicity.

The other lesson for me is I lost sight of my original intention, which was to redesign this diagram:

original

I do still think a Sankey could still work for this, with Q3 on the left and Q4 on the right, but because the instructions I found didn’t cater for change in volume over time, I just built a Sankey using dummy data.

So I proved that I could follow and adapt instructions, but in terms of the ultimate objective of data visualisation – to communicate – it was a hollow exercise.

 

The Friday project – week 8

I am a bit behind with the blog this week but I managed to add two more vizzes to my portfolio, and wanted to briefly highlight one element of each: the importance of preliminary sketches, and the importance of iteration.

Preliminary sketches

Makeover Monday this week used a detailed dataset on EU potato production, and I decided early on I wanted to take a more traditional, tiled, dashboard-style approach.

After some exploration of the data in its original form and in Tableau, I created a (very) rough sketch of my dashboard structure: an overall view (including tree map) in the top half , then the ability to drill down to see key metrics by country in the bottom half.

Here is the rough sketch and the finished viz side by side – it’s not going to win any prizes for draughtsmanship, but for a dashboard-style viz where a more disciplined approach is required, I find it an integral part of the development process.

 

Iteration, iteration, iteration

I also spent some time this week working on a viz showing the respective fortunes of Football League clubs over the last 16 seasons: The Ups and Downs of the 92.

Like the Makeover Monday dashboard, the main half of the viz shows the overall picture (a line for each of the 92 clubs, with the ability to filter by current league or club), with a more detailed view in the other half: in this case, individual graphs highlighting the respective fortunes of rival clubs.

the-ups-and-downs-of-the-92-club-rivalries-first-draft

 

But something was niggling away at me – the feeling that the top-left graph wasn’t telling the whole story. In the text I had referenced AFC Wimbledon’s six promotions in 13 seasons, but the graph only shows two promotions as the dataset I created didn’t go below the Conference.

Another iteration was required. I trawled Wikipedia for details of the previous seven seasons and added them to my dataset just for AFC Wimbledon. I wanted to keep the size of the axes in the right-hand graphs consistent to allow comparisons, so I dropped the bottom left graph.

It was more work, but the central story is much clearer for it: the steepness of the line and the inclusion of the additional four non-league segments brings home just how impressive AFC Wimbledon’s rise through the leagues has been.

the-ups-and-downs-of-the-92-club-rivalries

Tableau Public links:

Which EU-28 countries lead potato production?

The Ups and Downs of the 92

The Friday project – week 7

A really quick update today as I have been battling a heavy cold (just in time for my week off!)

Luckily the weather was so bad on Sunday that I had Makeover Monday in the bag before the week even started – here is the finished viz:

valentines-day

It is a simple viz but I feel that it brings out the two most interesting themes I found in the data: the unexpectedly large (for me – I must be a cheapskate!)  amount Americans spend, and the fact that they don’t just spend on their partners. Who knew!

The respective areas of the heart shapes are harder to compare than, say, a bar graph, but you can discern the more obvious trends, and the detail for each data point is available in the tooltips, including what percentage of people spent in each category.

And that is it for today – I did start compiling a dataset of English football league positions I hope to complete and use in the future, but it was a bit of a write-off as far as vizzing goes.

If I am feeling better I will probably do Makeover Monday early next week – after seeing the 100% club t-shirts for 2016 my aim is a full set for 2017!

 

The Friday project – week 6

A shorter blog today so I will start with my effort for Makeover Monday.

Cab usage in Chicago

Unfortunately I had to use the smaller dataset which only included 2015 and 2016 data for pick-ups from the Near North Side.

Here is the final viz below – my original idea was to create two: one showing the overall trends and one narrowing the focus to a specific angle, possibly cab usage on St Patrick’s Day.

In the end I only had time to show the overall view, and used a much simpler design than my original plan to showing multiple view of the data in a dashboard style.

I think it tells a clear story though, and it was another opportunity to experiment with design and colour in a way it would be hard to do in my day job.

I also learned how to add an info button which provides more information when you hover over it in Tableau Desktop, an approach I will definitely take again to keep my visualisations uncluttered.

cab-usage-in-chicago

Working on my CV

From Monday to Thursday I work for Lloyds Banking Group, and blog about Data Visualisation on their version of Jive.

The group has been a great way of connecting with people all across the Group who have an interest in Data Visualisation, and is the 25th most viewed out of over 4,000 groups.

One of the most popular threads has been on infographic CVs, and I spent the rest of the day turning mine into something a bit more visual.

Here is the result (I will update the featured visualisations as my portfolio improves) – let me know what you think!

data-visualisation-cv

February digest

ny_25_charts

Data Visualisation: The Stolen Art

Some top tips on how to integrate third party art into your own work in a respectful way.

banksy

How Alternative Facts Rewrite History

A reminder that alternative facts did not begin with Donald Trump.

ft_alternative_facts

One Angry Bird

A truly innovative way of exploring the emotions of the last ten US presidential inaugural addresses.

trump_obama

Hans Rosling: An Appreciation

And finally, in a sad week for the Data Viz world, a lovely tribute to the late, great Hans Rosling from Robert Kosara

rosling

The Friday project – week 5

A productive day today as I created two new vizzes and learnt some new skills: how to create a text box with conditional commentary, and how to customise marks on a chart.

Here are the links I used first up in case they are of interest:

http://kb.tableau.com/articles/howto/creating-dynamic-titles-based-on-filters

https://public.tableau.com/s/blog/2013/10/creating-and-utilizing-custom-shapes

And here are the final vizzes…

How much did UK clubs earn from Euro 2016?

euro-2016

This was my first attempt at taking Andy Kirk’s advice to “think like a data journalist”. I had seen this article the previous evening about the amount Ipswich Town had earned, and after some googling found a complete list of earnings for all UK clubs in this BBC article.

I found geodata for each club at https://www.doogal.co.uk/FootballStadiums.php as I was keen on using Tableau’s mapping functionality, as well as which leagues each club was in (with the exception of a few data points I had to source and enter manually).

The end result was almost exactly what I had sketched out the night before, with a selective bar chart on the left and the geographic view for all teams on the right, and more detail available through the tooltips, or on demand for each club via the dropdown.

Makeover Monday – Employment growth in G-7 countries

g-7

I had read Chris Love’s analysis on the data used in this week’s Makeover Monday earlier in the week, but in the end I decided to take the Andy Kriebel route and just make a better f*ing chart.

ak_tweet

I knew I wanted to learn how to use flags for the marks, but the question was: for what chart type? My original idea was a scatterplot, but this felt to me like I was over-emphasising the connection between the two metrics.

Ultimately I ended up using the circle views option, but with the two charts side by side and the countries in the same order on both, to allow comparison between the two metrics without trying to force a connection.

As with all vizzes, it is far from perfect, although in this case the dataset itself had significant limitations. Still, I had learned something new…and it was better than the original f*ing pie charts!

 

The Friday project – week 4

A slow start to the Friday project this week, as after a frustrating week at work I woke up as grumpy as Andy Kriebel if someone asks what Makeover Monday is.

Still, in theory the data was more in my comfort zone than last week’s tweets – tourist spend in New Zealand by district.

The final viz

nz-tourist-dollar

First thoughts

My first thoughts were that the use of the index was curious – it gives a sense of change over time for each district, but not the relative value of spend in each. This would clearly impact the final design.

Based on the original column charts, overall international tourist spend appeared to be rising and seasonal – in comparison, domestic tourist spend appeared to be more stable and more evenly distributed across the year.

It was time to explore the data in tableau, check my initial comprehension of the data, and understand the trends at district level.

Exploring the data

My next step was to build the graph I had scribbled down earlier, plotting overall domestic and international spend by month – as expected the international line was much more up and down than the domestic line (see below for the formatted equivalent), with an overall upwards trend.

nz-tourist-dollar-graph

I was pleased with the clear story  this chart choice showed and ended up using this with one fundamental enhancement – the ability to filter by district.

Looking at the district view generated some interesting questions – the largest outlier was Matamata-Piako district (see below). A quick google revealed why: the area was chosen to play The Shire in the interminable film adaptations of JRR Tolkein’s books for children.

mahatma

Design choices

I was happy with the chart I had chosen, but this led to a quandary – it only filled one third of my dashboard. I was keen to find a bold, appropriate image to fill the space, and after some google image searches (filtering by images labelled for reuse) came across the photograph below of Lake Matheson on the Wikimedia Commons site.

(For an excellent overview of the ethics of using other people’s work in your visualisations see Ryan Sleeper’s article Data Visualization: The Stolen Art)

Lake Matheson just after the sunset

Now I love this image (maybe because it reminds me of an area graph :-), and I soon decided that this should form the centrepiece of my visualisation. The dominant colours also determined by overall design – a black background, with grey and orange used to distinguish domestic and international spend.

And that was it. I added the fern design, partly to guide the eye to the dropdown, a suitable (if slightly boring) title and sub-title, and some annotation to add context to the numbers and some background to the remake.

Final thoughts

I am pretty happy with the result, and with my improvement since week 1, although counter-intuitively I ended up spending more time on this than last week. Still, there was still time to update my blog, and catch up on my Data Exploration and Storytelling course. And I was considerably less grumpy!

 

 

 

 

 

 

 

 

 

 

 

The Friday project – week 3

 

Week three of the Friday project, and instead of kicking the day off with Makeover Monday – I couldn’t face reading Donald Trump’s tweets before breakfast – I started with a blog post: Visualising football: title odds, shot dominance and winning streaks

Then, with a heavy heart, it was time to download the data, all of Trump’s tweets and retweets between June 2015 and December 2016.

Ninety minutes later, I had my finished viz, and some answers to questions I never thought I would ask…

trump_dashboard

Question 1 – where can I find images of Donald Trump’s hair

Why, http://trumpshair.com/ of course! Next?

Question 2 – what colour is Trump’s face?

Why, What Color Is Donald Trump’s Face? of course! (I love the internet).

Question 3 – how do I create Donald Trump’s mouth in tableau?

Right click on a bubble, then chose annotate > area. Size and reformat.

The final result is far from perfect of course (nor is it totally original – I modelled the mouth on the Time cover below).

 

For instance you can hover over each bubble to and read a tweet in the tableau public version, but not for the upper half of Trump’s face, where the hair image overlaps the packed bubble chart.

Over the course of the week, I had seen several beautiful and detailed reworkings of the same data, but ultimately I am happy to have ended up with something ridiculous and lacking detail.

Whether the US electorate is too, we will find out in the next four years.

 

 

Visualising football: title odds, shot dominance and winning streaks

An irregular round-up of visualisations on a single topic: this month, football!

Who will win the Premier league?

FiveThirtyEight is best known for its political coverage, but founder Nate Silver has a baseball background and the site regularly covers US sports.

Yesterday they launched a new interactive including team ratings, odds for upcoming matches and forecasts for the top 5 European domestic leagues and the Champions League.

There is lots to explore, but I wanted to compare the way FiveThirtyEight chose to visualise their Premier League predictions compared to the FT.com’s recent effort

(If you are interested in how the methodology of the two predictions differ, there is a good summary on John Burn-Murdoch’s twitter page).

I much prefer the FT’s version. Both visualisations are ordered by the expected final standings, but the FT version displays the relative strengths of the different teams much more intuitively, and the break from the visual norm is more eye-catching than FiveThirtyEight’s more traditional league table layout.

Shot dominance

Moving down a division, this graph which shows exactly why I have attended fewer Ipswich Town games this season than in the previous twenty.

experimental_361A scatterplot is the perfect choice here, with clubs divided into clear, descriptive quadrants, and interest to be found in comparing their position in the chart to their position in the league.

Winning streaks

The next visualisation by Jorge Camoes shows performance over an entire La Liga season (the original can be downloaded from http://www.dataatworkbook.com/data-work-11-change-over-time/)/

The good news is that this is actually a standard chart from Excel 2010 onwards – just go to Insert > Sparklines > Win / Loss (using 1, 0, -1 for W, D, L).

I do like the simplicity of this method – in his book, Camoes suggests using a dot or horizontal line for a draw, but even without this it is the Barcelona of Excel default graphs.

camoes_la_liga.png

The Guardian took a more creative route to show similar data, visualising Leicester’s winning streaks as shapes oddly reminiscent of a string of carrots – the interactivity is fun though!

guardian_leicesterWhich brings us back to the start – what odds would FiveThirtyEight or the FT have given Leicester last season before they pulled off the second most impressive rise to become champions of England since World War Two?

And yes, I do mean the second most impressive 🙂

538_leicester

 

January digest

A collection of Data Visualisation links from the last month.

A Datapoint Walks Into a Bar

An interesting talk from Lisa Charlotte Rost on how to evoke emotions with Data Visualisation (the sources are also well worth a look).

jd1

This is the most beautiful data visualization of all time, according to Reddit

I might agree if it was visualising my own baby’s sleep patterns – there is definitely a business opportunity here though!

jd2

Can tuna prices predict Japan’s GDP growth?

I have a weakness for charts that resemble the subject they represent…even if they cheat a bit with connecting lines.

jd3

You Draw It: What Got Better or Worse During Obama’s Presidency

Another great interactive visualisation from the New York Times – just draw your best estimate, then compare it to the actual trend.

jd4