I learnt alot with this challenge, but first here it is.
This seemed easy at first, Tableau supports date parsing with a function so after some research I applied the following formula but as you can see from the image it wasn’t parsing the 24 hour clock correctly.
A tweet and a few replies later, that was solved. The 24 hour clock needed “HH” rather than “hh”. I also picked up a valuable resource for the future, the date time format syntax dictionary here was pretty robust.
That said that wasn’t the end of it. The last thing to notice was that the tweets were downloaded with the GMT “+0000” time stamp rather than Donald Trump’s most likely region “-0500”. The final calculation below.
Rebuilding the visualisation
Apart from being painful, I learnt alot during the process. I lost the viz and wasn’t able to salvage my first attempt at this viz so I had to rebuild the whole thing. In doing so, I came at it slightly differently. Apart from saving often and backing up the file, on my second pass, I made the entire layout modular. Essentially because the layout / design didn’t vary from year to year, I built 2016 first ( also the phone layout) and then simply replicated the outputs and changed sets (used as filters) between sheets.
I really pushed hard on the design with this effort. That said the idea / look wasn’t original. I struggled finding an angle so I took to my viz library and settled on a design by Nicholas Felton. Whilst the data he used was different, the visual communication style suited the data and I also saw a challenge in trying to recreate that style in Tableau. I often hear developers from other tools or other technologies such as web design point out that you can’t control design as well as you can in other tools and whilst on some fronts I hear that criticism, it doesn’t rule it out altogether. You just have to make use of what’s there.
Here’s the 2013 Feltron report image I used for inspiration.