A crash course in a successful Data School application

The Data School application process requires you to complete two different data vizzes. You first analyse a Tableau Public sample dataset, before being given one if you get through to the final stage. Having become a Data Schooler myself as a part of its recent cohort DSAU11, here’s a quick crash course to give you the best chance at getting in!

 

Spend time with the data

 

Before even opening Tableau, spend time with your dataset. Open up that bad boy as an Excel and get acquainted with calculated fields, definitions, and what the data is trying to tell you. This stage also allows you to consider data cleaning and prepping tasks that may need to be undertaken before analysis.

Part of the feedback of my first viz was that it lacked a narrative — you run the risk of alienating yourself from the larger picture if you’re too gung-ho with Tableau. Spending time as early as possible researching the given topic online, as well as identifying any promising additional datasets, helps to make relevant connections and hypotheses. Breaks are a known facilitator of memory consolidation and creative thinking. With your newfound knowledge of your chosen topic, you may find yourself contemplating potential data stories in the time spent away from your project. This is where your light-bulb moment may occur!

 

Design for it to be understood

 

Having qualifications in Graphic Design and Psychology, this is an area I feel pretty passionate about. In essence, your final viz needs to be thoughtfully designed in an engaging way. There’s plenty of literature on the web that marry visual perception with data analytics, and to be honest, I find it hard to sum it up in a few sentences. So, here are some excellent talks with timestamps about infusing data analysis with design principles and fundamentals of visual perception.

 

 

Cole Nussbaumer Knaflic delivers an accessible talk about the basics of visual perception and storytelling before providing impactful examples and answering questions from the audience.

Cole covers visual attention and preattentive processing, data storytelling, and finish with a Q&A.

 

 

Heavy on the good stuff, Noah Iliinsky delivers a theory-laden talk about how we perceive the visualisation of data from a psychological perspective.

Noah covers patterns and pattern violation, intentional design and visual encoding.

 

 

A bite-sized TED Talk and a good introduction to the subject by David McCandless, the author of Information is Beautiful.

 

Yep, there’s a lot here, but listening to even a couple of minutes of any of these talks and you’ll realise the importance of presenting your viz in a thoughtful way. Good luck!

 

Structure your work

 

It’s easy to feel overwhelmed at the start of the project. I felt like either procrastinating until the last moment or spending every waking hour on my visualisation! In the end, loosely planning my viz helped to structure the project in a manageable way. Here are some general steps I used for my final visualisation.

 

Research

As explored earlier, research makes it easier to understand and link fields with the ultimate goal of a narrative-based presentation.

 

Data Preparation

Probably the least exciting part of your project but arguably the most important. Identifying null values, removing whitespaces, implementing correct data types and checking the quality of the data will help with analysis. If your data is wrong before working with it, your results will be too!

 

Exploration

By this stage, your data will be cleaned and ready to explore. Bring it into Tableau and start by creating as many graphs as your heart desires, with the goal to find any hidden stories that buck the trend. Your file might get pretty messy at this point!

 

Analysis

You’ll hopefully have found a story or two by this point, but not know the reasons why. Well, this is where the research that you conducted right at the start comes into play. An additional dataset, fancy analyses like ANOVAs and regressions if you’re into that kinda thing, or both, will give you the answer.

With my final dataset, a spike in the available seat kilometre measure per flight resulted from an additional dataset detailing fleet changes I identified earlier from planespotters.net. Although this stage can take the most time, it’s also pretty exciting.

 

Presentation

Breaking down the presentation of your viz verbally and visually, whilst being a pretty sweet alliteration, also helps with storytelling. They need to complement each other to tell a narrative.
Introducing the data, building up to the question of why before answering it, and then bringing it all together in a nice lil’ final slide will help with the ultimate goal of communication.

This isn’t a definitive guide as everyone works differently, but it’s important to know what you want to achieve, when you will achieve it by, and how.

 

So, there you have it. A (relatively) quick crash course guide on how to successfully apply for the Data School. Good luck!