To say that Week 3’s presentation was hard would be an understatement. The entire day flew by me as I struggled immensely with my selected Workout Wednesday, and I had to ask for a lot of help to even get started on the data. However, in saying that, I have learnt a lot (even if it wasn’t necessarily in the space of LODs or Table Calcs) and was instead a valuable lesson in how I can set myself up for failure before even starting on the visualising process.

Early Problems

The Workout Wednesday I chose was centered around re-creating this chart displaying some statistics on Week 7 of the 2020-21 Premier League season, and immediately I ran into a lot of problems. Firstly, I hadn’t realised that the data source had updated after the initial upload of the visualisation and now displayed the results of the entire season as opposed to the smaller sample size in the example. This led to a lot of panic and confusion for the initial hours of the day, with me wondering why every single calculation I made returned the same result. A more thorough check of the data source before hopping right into Tableau could’ve easily avoided this early trouble. Lesson learnt at the expense of almost 1.5-2 hours of work.

The problems continued, though creating the list of teams and the columns of tracking wins, losses, and draws were somewhat straightforward in the required calculations, and the provided information in the data source was extremely helpful in generating this data, I ran into an immediate roadblock with the final part of the exercise which was, strangely enough, the tooltip.

Issues with Data Formatting

In the actual challenge post, the designer of the problem wanted the tooltip of each of the last five matches to show the opponent of each match upon hovering. The problem was, that was physically impossible or extremely difficult given the method in how I created the data for all the teams. As the data source itself had two separate columns for the home and away team of any given match, I pivoted them together in order to actually create a new column from which I could get my team names without them being tied to a specific side of a game.

This was totally fine for the rest of the challenge, but also rendered my tooltip completely useless. As I had amalgamated every team into one column and their home or away status in another, I now had no way of being able to write a calculation to output the opponent of those last 5 matches each team played in. Perhaps there is a more convoluted way I could further transform the data using the time field of each match, but I suspect that would just undo the pivoting that I had performed and in turn broken the other charts.


The main takeaway from all this was that I had basically set myself up for failure at multiple points of the challenge but was taught valuable lessons about the process itself.

  1. The data being given will rarely be in a useable state that meets my needs. While the data source itself was detailed and well-documented, I still had to transform it to be able to even start working on it, something that I will no doubt encounter in the real world. Leading on from that, I should also be more vigilant in examining the data before any work. I was complacent due to the exercise nature of the task at hand and assumed that the data would already be correct to the listed specifications but should’ve verified this myself. In the future, this will most definitely be the first part of my work pipeline.
  2. Plan ahead. Had I bothered to actually check the tooltips and form at least a vague mental image of how I’d achieve it, I would have realised that the pivot method of combining my two columns together would make this an impossible or at the very least, extremely difficult task. Instead of being so eager to start visualising and creating charts, I should have sat down and mapped out my entire process so I had a clear endgoal rather than blindly stumbling around in the dark.

Overall, this was a very educational experience, though my own personal frustrations about the challenge being less about the aforementioned LODs or Table Calcs and more a challenge in preparing data for visualisation and analysis, it served as an early sobering reminder about what I may encounter within this job.

Daniel Yam
Author: Daniel Yam