Explanation and exploration with Tableau
This reading shows you the main steps and concerns raised during of the creation of the „Meteorites” Tableau visualization, which you can also download from Tableau Public Gallery.
Finding the right dataset
Looking for interesting data is always challenging and it gives you new possibilities to view something different in a different way that you used to. You can find many open datasets, especially from the U.S. and the U.K. since these countries are far ahead realizing the open data initiative. There are also other freely accessible datasets where enthusiasts like you collect and share data with each other.
When you are about to find a public dataset in the midst of the enourmos information galaxy of the internet, you have to face a few questions that will verify if it worth playing with that particular table or not.
These include, but not limited to:
– What extent does your set contain data?
– Is the granularity of the data adequate?
– Are you interested in digging into the data?
– Is the source of the data reliable?
– What other sources can you have?
– What other trails can you pick up from the dataset?
… and so on.
Here’s what I’ve found and used:
- Dataset: http://visualizing.org/datasets/meteorite-landings
- Classification: http://www.meteoritemarket.com/type.htm
- Other information: http://www.wikipedia.org/Meteorite
OK, we have the meteorites dataset, it seems that we can work with it, let’s see how to prepare the data and the worksheets!
What does your data look like?
After you downloaded the promising dataset, take your time to inspect it, get to know it, see the structure of the data, feel it inside your nervous system. This is really important since you’re going to work with it, so you really need to know what it contains, even if it’s just by glance.
A little data quality management
When you work with data from exotic sources you need to be sure that the set meets certain quality standards and it is standardized in a way so you can use it for your visualization. First, I suggest to eliminate those rows that contain null values, so later on your way you won’t be bothered by errors you can not see where are from. This is just a playground advise, of course if you work for a client you shouldn’t just throw out the null values/rows because it is convenient, but in our situation it is allowed.
Loading data into Tableau
When you clean up your data you can import it to Tableau which gives you many types of different connectors to different data sources, so you don’t have to worry about it if you accidentally used a VectorWise database server. Seriously, when you load your data you need to be aware of a few things that can easily get you into trouble. These include but not limited to things like picking the right field separator. Try not to use comma or semicolon, because if your data contains text than you might have a problem. Better choices are the characters that don’t usually appaer in datasets, like the vertical bar. You have to be aware of the character coding as well, since all those data manipulation apps produce different character encoding (UTF-8 is the best in my opinion) which can generate interesting results. When your data is loaded successfully you should see your measure and dimension names to appear on the left shelf of Tableau.
It is much advised to design your visualization before you start creating it. It is much easier to change something on a paper than in Tableau. At least a draft please. Paper prototyping is an exciting way of creating drafts and wireframes not only for your visualization but also for other projects.
Typography and grid
A good typo can give you a visual boost if you apply it on a corresponding grid. You can find many books on how to choose a certain family of type for a certain content and how to create a grid that holds up and structures your main content. A grid can help you easily rearrange your elements (if you haven’t made a paper prototype already) and can give you pixel perfect positions with a balanced view. Who said there is no pixel perfection in Tableau?
Creating the worksheets
When I started to create my visualization I was not planned as detailed as I should have, only the main topics/aspects were on paper front of me to guide the construction. But it was fair enough to bright those sparkles.
Collateral images (and damages)
An image worth a thousand words – as the famous saying goes. But which thousand? Choose your images wisely since they can create more confusion than clarification sometimes.
Editing the texts
Yes, you will have to do it. When you create a complex visualization that contains graphs, symbols and images, you probably will incorporate some text that introduces or adds some textual reference to your subject. C’mon, you will need headings at least! And yes, you have to do it. Pay close attention to the labels since these will be the guides for your readers about what they are looking at when viewing your beautiful visualization. Try to use the language of your target audience, and not to use the words of some rare scientific subfield.
When you use an image or a piece of text to help your visualization that is not your own creation you must credit the original author. This relates of course to the dataset as well! There are different types of credits that you can use, but these must be in line with the original author’s request.
Click here to view the entire viz and don’t forget to turn on full screen!
Latest posts by Tamás Földi (see all)
- Scaling out Tableau Extracts – Building a distributed, multi-node MPP Hyper Cluster - August 11, 2019
- Python Experiments in Tableau 1. – Add live currency conversion to Tableau Dashboards using TabPy - January 9, 2018
- HOWTO: Tableau Server Linux in Docker Container - November 2, 2017