A Redesign of a Redesign
I would like to start out this put up by stating that I’m an enormous fan of Cole Nussbaumer’s work at Storytellingwithdata.com. In case you are not aware of her information visualization weblog then I’d extremely suggest that you just comply with her. Cole affords sensible recommendation for designing nice information visualizations and sometimes has fantastic redesign examples on her web site. With that stated, I wish to look at considered one of her redesigns that I got here throughout in the present day and supply another strategy.
The unique information visualization and her redesign notes are here. That is the unique information visualization:
Right here is Cole’s redesign:
She was gracious sufficient to put up the Excel file that she used to create her redesign. It is out there here if you’re excited about exploring additional.
Level #1: There’s a information error in her redesign.
For those who look at the unique visualization intently you will notice that Competitor 5 leads in Class 1 and is in final place in Class #2. Nonetheless, in Cole’s redesign this isn’t the case. It’s because it seems that Cole made an harmless mistake when copying and pasting values from row to row in Microsoft Excel when she rescaled the information. Class 1 and 6 are literally the one ones which might be appropriately assigned.
In her redesign:
Class 2 is definitely Class 3
Class 3 is definitely Class 5
Class 4 is definitely Class 2
Class 5 is definitely Class 4
All of us make errors. I’ve actually made a lot of them through the years. The one recommendation I can give you on this space is 1.) all the time test your work and a pair of.) if doable get further eyes to look at your closing product. It is not a foolproof answer, however it would aid you catch errors like this. This error is hidden pretty effectively. It’s onerous to right away acknowledge due to the best way shade is getting used with the bar chart. This results in the subsequent concern that I’ve with this redesign.
Level #2: It is vitally onerous to find out one competitor from one other.
Cole makes use of a quite simple shade palette, a blue-gray shade palette that she makes use of typically. It is a very pleasing palette and is non-distracting. Nonetheless, as a result of the entire opponents are grey she makes use of the order of the bars to differentiate one competitor from one other. This requires the attention to travel to the legend to find out the order after which rely down the bars to find out which competitor is which. Consequently it is vitally troublesome to shortly decide which competitor has the lead, or any given place, in every class.
Level #3: The size has been arbitrarily rescaled.
Reworking variables is an ordinary apply within the subject of knowledge analytics and whereas I like her idea, this rescaling might trigger great confusion in sure circumstances. The unique visualization has the information factors labelled as « weighted efficiency index ». This appears very particular and is probably going calculated by the enterprise in a sure means. Cole has rescaled this to a constructive scale by including 1.1 to each quantity. Nonetheless, there isn’t a indication in her visualization that it was rescaled, there isn’t a axis to supply a zero bottom line or the vary of the values and she or he makes use of the identical label of « weighted efficiency index » as the unique visualization. This has the potential to trigger nice confusion in a real-world software. For instance, what would occur if readers of the report have been used to seeing detrimental values for the « weighted efficiency index »? There can be numerous questions. As well as, any comparisons to earlier studies can be misplaced. I notice that if this have been a brand new report or the size is not an ordinary enterprise metric that it might not be a difficulty, however I would like a redesign answer that doesn’t rescale the information.
Level #4: The rating would not inform the entire story.
I actually just like the rating thought to simply evaluate « our enterprise » to the opponents. Nonetheless, it would not inform the entire story. For instance, contemplate Class 1. On this class Competitor 1 and « our enterprise » have the identical worth. It is a widespread downside with pressured rankings. There’s a tie for fifth place and a choice must be made on the right way to deal with that order. On this case Cole made the selection to place « our firm » in final place.
There are different nuances which might be additionally misplaced with the pressured rating system. For instance, in Class 5 (Class 4 in Cole’s redesign) « our enterprise » ranks third place, however the efficiency on this class is far larger than the efficiency in Class 3 (Class 5 in Cole’s redesign) the place it ranks 2nd place. In different phrases, we overshadow the truth that 0.51 is larger than 0.19 as a result of rating turns into the point of interest of the comparability. The relative efficiency in every class is misplaced. Ideally we might see each, the rating inside a single class and the relative comparability throughout classes.
Take into account the opposite vital questions that can seemingly be requested:
Who’s the best-in-class of all of the opponents? i.e. the largest competitors and who’s the worst?
Is that this straightforward to find out from the visualization?
Level #5: Is that this a bar chart or a rotated histogram?
Bar charts are used for categorical comparability and histograms are used to indicate distributions. The final steering on bar charts is to have some area between the bars. There’s a superb line between an excessive amount of area and too little area. Histograms normally have a smaller area to point a relationship of the distribution.
Sadly there’s not a lot analysis on this space, however there are some common pointers that specialists recommend. For instance, Stephen Few addresses this in his e-book Present Me the Numbers in addition to a put up on his weblog Perceptual Edge here. Even utilizing a minimal area on this case can be higher than no area in any respect. With none area between the bars the bars come collectively in a big block and will it very troublesome to learn.
A few of these factors may very well be be resolved by making just a few minor adjustments to Cole’s visualization. For instance, on this redesign I saved the size the identical as the unique and added somewhat area between the bars to make them simpler to learn.
This does not clear up level 2, i.e. the problem of figuring out the competitor shortly. Utilizing shade is an alternative choice, for instance utilizing a categorical or sequential shade scheme to double encode the competitor with the bar order. This instance makes use of a sequential grey to assist differentiate the competitor.
This is not best and I can see why Cole averted this in her redesign. Shade turns into distracting and utilizing a categorical shade scheme is even worse. The opposite downside that’s now launched with this strategy is that the reader is now pressured to match the lengths of the bars in two completely different instructions if they’re making an attempt to make actual comparisons.
My closing redesign makes use of the unique idea of a dot plot. Within the authentic design completely different shapes and colours have been used to distinguish the opponents. Utilizing shapes aside from circles or dots shouldn’t be best, however I’ll save that detailed dialogue for an additional put up. The usage of shade within the authentic can be not good as a result of they’re mixing vibrant alerting colours of yellow and orange with extra delicate blues and purple.
Beneath is my redesign accomplished in Tableau. Click on here for the Tableau Public workbook.
Through the use of a easy dashboard highlighting motion in Tableau it makes it even simpler to look at every competitor intimately as effectively. If making a static report then small multiples may very well be used to make fast comparisons. It is now very straightforward to see that Competitor #4 is the most effective general and Opponents #2 and #3 are the 2 worst.
I hope you discovered this redesign dialogue fascinating and ensure to comply with Cole’s weblog!. As all the time, when you have any questions be happy to e mail me at Jeff@DataPlusScience.com
Jeffrey A. Shaffer
Observe on Twitter @HighVizAbility