It’s the final day of dashboard week!

Today there was no particular challenge – just make a dashboard! Our data was the Melbourne Census of Land Use and Employment (CLUE) data.

And wow, is this data big! It has land use and demographic data from over 10000 properties across 17 years. Given one day, grinding this data down to one dashboard was a significant challenge.

Before I go into any details, you can find the dashboard here. I have included a screenshot below, but I would encourage you to view it on tableau public and click around – the main purpose of the dashboard is its interactivity.

Dealing With The Data

My approach today was to immediately simplify the data – the CLUE data itself was too big for me to understand in one day. Especially for Sydneysider like me!

So, instead of exploring the data I narrowed myself down to only looking at one particular aggregate file we were given – this showed various categories values for each block. This included residential and commercial carparks, cafĂ© housing and student accommodation. My goal became to build a dashboard that describes these categories’ relations to one another. The dashboard allows the user to explore which blocks and regions occur for which function – which neighbourhoods are primarily residential, industrial or commercial?

With this in mind I designed a drill-down dashboard. It has two steps – click the correlation heatmap to show a map drill-down. Then click the map drill-down to show a line chart over time. In this way the user can explore relationships between categories in detail. The dashboard itself is technically simple; all of the charts are basic chart types in Tableau. There are a lot of dashboard actions – it was important for me to name these actions descriptively so I could manage if any of the actions errored.

Reflections on Dashboard Week

Now that dashboard week is over, I have time to reflect on the process as a whole. It has been undoubtedly hectic – building a dashboard a day (with additional challenges) is a lot of work over one week. Yet I feel that my biggest success this week has been in timeboxing – I did not stay up late to do any dashboards. Even when the resulting dashboard was less detailed or a lower quality than I expected, I was able to put out a product that I was proud of (see Tuesday’s blog).

This week I was also able to create a variety of dashboards of differing styles. Monday’s, Wednesday’s and Friday’s dashboards were more serious and informative. Tuesday’s and Thursday’s dashboards were more fun and creative, pushing the boundaries of Tableau’s capabilities and taking an unexpected approach to the dataset. This kept things interesting for me throughout the week (and kept me sane).

Perhaps my biggest takeaway is that I am looking forward to a restful Friday night – this blog might be goodbye for a little while as I take a break from exploring my dashboards for a little while!

The Data School
Author: The Data School