Today we had the task to perform some football analytics. We had access to json data parsed out in Snowflake, 36 tables in total, however we were advised to narrow our focus.
The 36 tables varied from league data, to data on the UEFA 2022 Euros and on specific areas of the game such as substitutions & goals.
After exploring the data within Snowflake, I was drawn to exploring the fouls so decided to use the table on fouls committed. I initially thought to use the bad behaviour data too, which had data on cards given. But I discovered the Foul committed table had card data in it too specific to the fouls logged so it made sense to use that one table.
The table contained data on UEFA Women’s Euros 2022 data where each Event type is a foul. There was an event ID but I discovered this was not unique and therefore made a unique key which was a combination of the match ID and event ID. The granularity of the data was each row was a foul event for a particular match. The table contained data for 31 matches.
Sketch:
I started listing out the data I’d like to look at & what questions I wanted to ask of the data. Then I added which columns I’d need to do this.
Following this I sketched out how I would visualise answers to those questions or charts I’d use to explore.
I talked through my sketch with Carl and crossed out some of the visualisations as I had included a lot, of which many were very much summary data.
Data Prep:
I was fretting about connecting to Snowflake in Tableau so got some reassurance on that from Carl!
Then I did connect to the Snowflake in Tableau Prep also but for more of a sense check on figures following some initial queries I had done in Snowflake.
I did not actually need to do any Data Prep in advance to starting.
The Dashboard build:
Once connected to the table in Tableau, I created an extract to help with the load time.
I began with my KPIs. Then I had to create some Fixed LODs for the number of fouls and cards per match for my scatter plot. Finally I needed to get the number of fouls before the first yellow card for each match, also for my scatter plot and an additional KPI. With Jenny’s help I had a running total table calc restarting each match and intended to count the zeros (numbers were 1 onwards from the first yellow card). However I ran into non aggregated/aggregated issues and I also wanted to be able to use the calculation outside of the table and was struggling with this. So with the help of Ollie and Ollie I got a series of calculations working to get this.
I wanted to add viz in tooltips for some of the KPIs to add further context (top 5 teams for fouls etc.) but was unable to add this in the allotted time.
I decided to take a stop at a certain time in order to give myself enough time to format the Dashboard, tidy the other tooltips, and add the Dashboard action for interactivity between my scatter plot and histogram.
Feedback:
Following my presentation of my sketch and Dashboard I received the following feedback:
- Add context to KPIs
- 15 min buckets rather than 10 (as presently there is a time bin of 40 - 50 minutes which does not tie in to 45 minute halves)
- Group the over 90 mins on the histogram
- It would be good t have more info on matches for the scatter plot rather than just Match ID (i.e. who was playing)
Next Steps:
My next steps are to take the above feedback and amend, and then add the viz in tooltips for the KPIs.
Challenges:
Overall I found this a great task to start off our Dashboard Week. I did run into challenges with some calculations which I was grateful to receive help with, but I think if I had stopped and planned out the steps I would have been able to solve it. It was the first time I had connected to Snowflake in Tableau working independently from others and I think this threw me even though it is very simple! So for our next challenge, I will remember to stop, slow down and give it a go even if I am not sure.
