A few years ago I started baking my own Neapolitan style pizzas. Despite its reputation, I learned that pizza is not a fast food. Far from it. The dough for a proper pizza base requires at least two days of fermentation (although you can get away with one day in a pinch). I've even made pizza dough that fermented for five days, resulting in incredible strength and taste.
A key element was the pizza stone that I got as a present from my girlfriend. Without it I could never reach the required temperatures with my conventional home oven. Pizzas would come out tough and chewy. But with a thoroughly pre-heated pizza stone I can get my pizzas done in about six minutes. A real pizza oven can do it in three. This is what gives Neapolitan pizzas their crispy charred crust that is still soft inside.
One of my favorite pizzas is the classic pizza Margherita. The pure flavors of tomato, mozzarella, and basil really showcase the quality of the base. Nevertheless I am not a pizza purist. I love coming up with interesting pizza toppings that are not traditional. Perhaps my best pizza is one with mushrooms and onions sauteed in white wine.
Pizza and Data
When Maven Analytics launched their 'Pizza Challenge' in October 2022, I couldn't resist participating. It combined two of my passions: data and cooking. The challenge was to take a dataset containing the 2015 operations of a fictional pizza restaurant, and turn it into a dashboard that showed the possibilities for improvement in the business. The original dataset can be found here.
The dataset contained a year worth of transactions, as well as some details about the pizzas sold. The data was mostly clean, but some of it was in a hard to use format. For example, the ingredients used by the restaurant were contained in a long text string for each pizza. I had to separate them and create a relational table in SQL to understand their use in various pizza types.
There was also a lack of exogeneous data to explain patterns in the performance of the restaurant, which would have made the analysis more interesting. However, because all the orders were precisely timestamped, there was plenty of opportunity to analyze activity over time. Making some assumptions, I was able to get a detailed insight into the capacity utilization of the restaurant, where there was clear room for improvement.
Visualizing the data
For the presentation of my data I decided to focus on three things:
- The seating and table occupancy throughout the week.
- The variation in demand throughout the day, week, and year.
- The performance of different pizzas compared to their ingredient usage.
To visualize this in a dashboard I used Tableau. In retrospect this may have been a poor choice. Tableau's strength is the ability to drill down into variables and apply different filters. However, this dataset didn't have many variables to drill down into or filter on.
Tableau's weakness is that it is clunky when you want to structure and arrange elements on a dashboard, or when you want to customize charts. It took unnecessarily long to get the elements of the dashboard aligned, although eventually I'm happy with the result.
Below is the dashboard that I ended up submitting for the challenge. You can also find a live version of the dashboard on my Tableau Public page. I've made some small cosmetic modifications to that version compared to the version below.
