To get into the Data School, you’re asked to create a Tableau visualisation using a provided dataset and present it during the interview, explaining your story and key insights that you uncovered. Now with my four months of training coming to an end, I wanted to take a step back and revisit the visualisation I created for that final interview. It’s a chance to reflect on what I’d do differently now as opposed to where I was four months ago.
We were given a dataset on the Nepalese Himalayan Mountains. I decided to look at mountains that are exclusively over 8000 metres and assess how dangerous the death zone is. The viz can be found Here.
Big Number Formatting
The big numbers at the start of the viz provide great information such as total expeditions, total climbers, total summits, and summit success rate. With the capability to filter by mountain, they offer further details for each specific peak. However, while the insights themselves are strong, the formatting can be greatly improved. Since they are ‘Big Numbers’, I would definitely increase their size to make them for of a focal point. Additionally, not sure with what I was trying to do with the floating containers being used as line separators between the BANs. They are different sizes, and the middle container is even a slightly different shade of blue compared to the others. I would remove these floating containers to clean up the design and ensure the spacing between the BANs are consistent.
Misleading Graph
Within the Timing the Ascent Section. There is a bar graph that shows expedition fatality ratio by summit time. While it looks interesting at first, it can be misleading since it’s based on expeditions and not individual climbers. For example, imagine there are three expeditions with eight climbers each. If there is one fatality in each expedition, the chart would show a 100% fatality rate. When in reality, it’s much lower on a per climber basis. I would switch this graph to climbers to give a more accurate picture of the risk.
Donut Charts
There are two donut charts, one comparing the fatality rate by elevation, and fatalities by season. These charts would be much more effective as bar charts. Since they are comparing more than two values, bar charts would be much easier to read and compare. Providing clearer visual distinctions and making it easier to interpret at a glance. I believe I originally used donut charts to mix things up a bit, since the viz was already bar chart heavy. But in hindsight, it’s better to stick to visualisation best practices and choose the right chart for the job.
In Summary
Most of the changes I would make are formatting related. Such as increasing big number sizing, removing the unnecessary separator line, and updating the title. I’d also adjust one of the calculations to look at fatalities on a per-climber basis instead of per expedition, to provide a more accurate picture. Lastly, I’d make sure to stick to visualisation best practices by using the right chart types for the data.
