I’ve come to The Information Lab with zero expectations. I knew I was going to learn Tableau & Alteryx from the brightest in the biz but that’s about it.
To sum up, the first week was challenging and overwhelming but also liberating and creative. It was a lot to soak in over 4 days so let me break it down.
(TL;DR : Feeling way in over my head until just the last moment, then I had a paradigm shift and breakthrough analysis. This reminded me why I am here. I am so lucky to learn things everyday from data and amazing people. Week ended with a celebratory beer.)
Challenging and Overwhelming:
We’re pretty fortunate that TIL is now home to 2 of the world’s 20 Tableau Zen Masters (with patience and kindness to match their brilliance). We got a chance to see some advanced Tableau work by them through TIL’s Zen Master Training. But in all honesty, I was lost and definitely discouraged by it. Followed by the practice Tableau Desktop Qualified Associate exam on Day 3 and working on the weekly assignment, I had some questions on how I was going to keep up or if I was even in the right place.
At the end of Thursday, I had literally scrapped everything that I had worked on. I had hit walls in connecting to data sources, missing data, questions about how the data was constructed– all which made implementing my hypothesis and line of questioning useless. Did I mention I present tomorrow?
Then Andy mentioned, “Why don’t you change the dimensions of the scatterplot?”.
Now this is kind of an obvious point– where I didn’t even think it was worth mentioning. Yet, with the the scrolling functionality at the top, I didn’t realize the viz’s UI reinforced the it’s argument so well that it made me feel like I had to dig for data to refute it. That’s one of the most important things I will learn in the next 4 months – that design is just as important as the data for the story.
By realizing I could breakdown the data into regions and using cluster analysis, I was able to tell my story – that spending is not the ultimate solution to increase uptake of school catered lunches for students. The real end-game is to keep hunger from interfering from a child’s right to education, regardless if that’s provided by the school or parent. With such large variables and potentially thousands of public funding dollars on the line, we needed confidence in this model. By doing a linear regression and cluster analysis, I found a couple of interesting points.
1) The South East and South West clusters have below average uptake and can learn from the North West where they have above average uptake.
VS 
2) The closest cluster confirming the relationship between catering spending per pupil with average uptake is London. It’s got p-values below 0.05 with an R2 of o.3. Its pretty reflective of the UK but probably because it’s neighbourhoods are some of the most diverse in the UK.
Liberating and Creative:
Once I got into the cluster analysis, I felt this white-hot sense of inspiration. I literally felt a sense of relief that there was a good story here. I got Andy to take a quick look at it in the morning and I learned a lot from how he interacted with my data. He quickly analyzed and made decisions about which storylines to pursue, weighed the pro/cons of finding more data to create a representation of the viz against impactfulness and time to our presentation. Watching him go through this process made me feel more confident about the sense of expectation here. Additionally, the crayon exercises were really helpful in reinforcing the idea that data presentation lives in the world of first impressions. Funny enough, the assignment gets harder the longer you work on it because its easy to forget the original goal. Its interesting to note that he personally limits himself to 1 hour on his Monday Makeovers.
Going back to the analysis process, I personally identify with Christian’s Chabot‘s view that easier ways to get insights is not only a business mandate, but a workplace liberation. I love to work with data but will be the first to admit, I’m not the best with the technical stuff. Once I got to these conclusions, tried to clean up the representations and front load my conclusions in this matrix, I was so excited to present. Getting here on the first project was pretty rewarding 🙂
Thanks for reading. More vizzes on the way.