On Friday DS14 were given a task at 9:00AM to create a dashboard via an external database, Exasol. The task was to sketch our dashboard on a piece of paper by only looking at the fields in the metadata (as we were not allowed to interact with the data beforehand), create a dashboard on Tableau and then present it after lunch.
That was my first experience to deal with a dataset of more than 230M records. One of the things I learned that day is that you can right click on any field and select describe at the bottom of the dropdown, which will basically display a summary of that field and what exactly it contains – as shown below:
So, my data set was about Ozone concentration measurements and after having an idea of what the field contains, I decided to focus only on the US, create a heat map to show when those ozone measurements increased and also show the highest 10 states by ozone measurements. This is what I sketched before interacting with the data:
Also, my dataset had some spatial data and being a newbie in dealing with such a huge size dataset, I was prompted to practice what we learned with Andy that week and use a hex-file created by Joshua Milligan in order to visualise those measurements across the US in a hex-map. I tried to join the spatial file but unfortunately, I lost 15 mins waiting for Tableau to only retrieve 30K rows which was a lot of time – given that we only had 3 hours to draw our sketch, create the dashboard on Tableau and then prepare to present it. So, there was noway to wait for Tableau to retrieve 230M rows, so I decided to move on and use Tableau’s default light map.
One of the most important lessons I learned from my first 2 projects at The Data School is to let go of some ideas when you get stuck however tempting they are – especially if you have to deliver within a short deadline and you can’t find a way to implement those ideas (or maybe you can – but it will take much of your short time).
Here is my final outcome:
It was really a great experience to plan our dashboard ahead and sketch it without interacting with the dataset. However, I realised that I don’t give enough time to practice my presentation and that I keep working on Tableau 90% of my time and 10% researching the topic to have some context, which I am planning to work on in my future projects.