#FastFriday – 6 Minute Vizzes – 1st Episode

Welcome to the first in the #FastFriday series.

For context, this is a weekly challenge that Simona Loffredo, Ravi Mistry and myself will be participating in. The concept is as follows – We have to design a visualisation created from a dataset which we see for the first time when we connect to Tableau. We’re starting with a time limit of SIX (6) minutes to create a visualisation with meaning.

The objective? To learn about interpreting data better. To understand each other’s approach when faced with a new data set. And too see how well we viz under pressure.

We will then screen record and blog our experiences.


 

Lets start with the recording of my 1st experience.


Understanding the data

First things first, if you have never seen a dataset before then you should always gain an understanding of the data. What it entails, the shape of the data, what dimensions are there? What measures are there?

In this case I used Tableaus data source pane to aid me in this part of the process. It quickly tells you what the data looks like and the data type.

In the case of this work I identified very quickly that we had a series of geographical dimensions, presumably in some sort of hierarchy, some summary measures for each dimension, followed by some more granular measures focussing on scores for certain aspects; each of these scores also had a rank field.

At this point I knew what I was going to do. I was going to guide my users to an understanding of which geographical areas have the what ranks for each of these aspects. ‘Who has it best’ for lack of a better name.

Now in future I am more likely to open a sheet for this exploration phase. The data source pane caused some time loss in that I had to drag columns across so I could completely see headers and also scroll back and forth when investigating data types. The data pane on the sheet offers me the same ability but without the limitations noted above. I can also Investigate the data if I wish by then quickly dragging on measures and dimensions that I may not understand.

Data Source pane


Building out my dashboard

In all my visualisations I add borders in order to guide my users to where I want them to look. I want them to direct their attention to within the borders. This is the first thing I did (although in this case, I did this really poorly and had to rework these several times).

I built this…

Poor borders

When I actually wanted this…

Good borders

I then added my title. I did this at this point to try and engrave the idea in my head. As if I thought that without adding the title at this point meant that there was still potential for flexibility if I found something different. I was trying to ensure there was clarity in the next phases of my work.


What data should I use?

One of my key design decisions at this point was specifically what dimensions and measures I wanted to use.

In terms of a geographical zone I chose to use the Local Authority field. This was because in comparison to the other geographical dimensions I felt this was likely to be the most recognizable to my audience.

Now, not having the most in-depth knowledge of a dataset I knew there could be a lot of variation in the traditional ‘score’ measures; I had no knowledge of what the range of values could be, which could of course cause problems if you have extremes that would cause some of the data to be ‘drowned out’. So I chose to work with the rankings, which were all likely to have the same range and have a linier increment.

Why no map you are probably asking? Well tableau did not appear to recognise my dimensions as geographical fields and instead string fields. This immediately tells me that perhaps if I were to try and build a map there would likely be teething issues which when trying to build a visualisation at high speed is definitely not something you need.


Choosing a viz type

It just so happens that the day prior to recording this visualisation I had been in the audience for a talk with Andy Cotgreave an evangelist for Tableau. In this talk, amongst other wise words, he discussed the merits of bump charts and heat maps.

Bump charts are designed specifically for ranking data, but for this use case I did not think they would fit. Bump charts are usually not just focussed on ranking by any dimension, but ranking by a time/date dimension. Thus having the joining line makes sense, showing a trend. They are also not a natural chart type in Tableau so can be a bit complex to sort.

There are also a large amount of dimension members that may make it difficult to track and refer back to what local authority they belong to.

So I decided to go with a heat map. They are incredibly easy to build in Tableau and are a great way of showing comparisons between categorical data. They provide your users with a visual clue about what values are high/low, they just have to look for darker and lighter colours.

This example from the Office for National Statistics is a great example and one of my favourite vizzes. It is so easy to identify acknowledge different trends within the dataset.

Heatmap

After building the heatmap I then had to make a key design decision, do I label them. I answered no, I thought that they cluttered the visualisation and the users always have the hover action if they wish to get this additional information.

But It is amazing at how the drag and drop nature of Tableau allowed me to  try this out and make a decision on this so quickly.

dsfg.png


Final design tweaks

Prior to completing the visualisation I made a couple of design tweeks that I felt were necessary.

Firstly I made the heat map calumns wider, this was to ensure the readers could see the entire headers for each of the different values.

Secondly I added a sub-title, this was to add a bit more context to the visualisation for my readers, allowing them to better understand what it is designed to do.

You may have noticed when creating my sub-title that I used a neat little trick to add spacing between the words. I always do this as I feel the default spacing Tableau offers is just a bit tight.

In order to do this firstly type in some random letters, or if your abit more organised something like ‘line space’, I went for ‘SDF’.

I then resized the font to the smallest size, 4 (not available via the drop down list so you must type that in), I then hit ok/return to submit the size of the font, otherwise this trick will not work.

You can then go back into your title box to and match the font of the text to your background colour.

Edit text

edittext2

edittext3


Post 6 minute troubleshooting.

So, after completing the 6 minutes I wanted to review any areas where I had made mistakes just so I could learn for future occurrence’s when working with similar data.

Firstly, I am using the rankings for a dimension that is higher than the lowest level. Thus I have not used the correct aggregation method by summing the variables. If a local authority has more ‘data zones’ that sit beneath it, then the sum is likely to be higher even though on average (what I should have done) the ranking for that local authority may be better.

Secondly, the view requires sideways scrolling. Wherever possible I always try to avoid scrolling in any direction. I should have been more selective in the measures that I chose to include.

The ‘Overall  Rank’ should have been included as either the 1st or last column. Our audience expect summaries to be at the start or end, by including it in the middle the audience are unlikely to acknowledge that this is a summary measure.


 

Summary.

I cannot explain how much of a valuable exercise this has been, even to do it just this once so far.

When watching the videos that Simona, Ravi and I have produced for this initial 6 minute viz, you can see the different ways in which we go about designing our visualisations, you can also see the little differences we make in our priorities.

For example in mine where I add a title at the beginning, and in Simona’s where she renames her calculations despite the time pressure.

This was a thoroughly interesting exercise and I look forward to see how I develop as a result of this series.

It would be mean of me to end on that point without giving you the chance to take part, so here I have attached a blind dataset (unlucky for you guys, not the same used by myself, Ravi and Simona).

We would love to see the finished output so please send us your completed 6 minute visualisations, and please screen record if you can!

One final thought, It is amazing that despite only having 6 minutes, you can still clearly see that the process flow remains similar to that of longer projects, just in a compact amount of time.

We have the initial uncertainty phase; exploring the data, looking for patterns or insights that can be gained, before clarifying an idea and focussing on producing the desired output that will best tell the story.

My 6 minute viz

Ravis #FastFriday blog

Simonas #FastFriday blog

uncertaity.jpg

Advertisements

2 thoughts on “#FastFriday – 6 Minute Vizzes – 1st Episode

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s