It’s Dashboard Week this week in the Data School. Each day this week I’ll be writing a reflective blog on the lessons I’ve learnt from having to prepare and visualise data at speed.


Today was a mixed success, but I’ve learned a few things.


  1. Narrow the scope of your analysis before you prepare the data. Today’s data set, the US Housing Survey, had tens of thousands of rows, each with hundreds of columns. Any workflow with multiple join and transpose tools runs slowly, at least when the deadline was only one day. Next time I’ll limit the amount of data before I start.
  2. Make sure you only transpose the fields you want to use as measures in Tableau, and leave the fields you want to use as dimensions / filters in columns. This makes it easier to analyse the data, as it limits the number of calculations you need to build.
  3. It doesn’t matter if all you’ve got are a couple of bar charts. Much of the work in creating an effective visualisation is in taking a simple chart and using colour, font and spacing to bring it to life, like so:





Admittedly, it still needs some formatting work (the viz isn’t well-aligned enough as far as I’m concerned), but it’s better than before.