BW #28: Pret a Manger

If you've ever been to London, then you've likely seen one of the many, many Pret a Manger sandwich shops. It's kind of like seeing a Starbucks Coffee in New York; they're just about everywhere. There are actually 446 Pret a Manger shops in the UK, and plenty of them are located outside of the busiest areas of London -- in suburbs, train stations, airports, in cities such as Manchester, and even in Scotland.

The UK's Office for National Statistics (https://www.ons.gov.uk/) noticed this, and realized that Pret shops might be able to provide some metrics on consumer behavior. (Or "behaviour," as they would have put it.) If sales are uniformly up across all locations, that might point to increased consumer confidence and growth. If they're all down, that might point to belt-tightening and problems. More interesting, though, would be if some went up and others went down, allowing for a rough understanding of what's happening across different sectors and locations of the economy in close to real time.

And indeed, the Pret data is part of an overall attempt to measure the UK's economy in real time. Every week (latest issue is at https://www.ons.gov.uk/economy/economicoutputandproductivity/output/bulletins/economicactivityandsocialchangeintheukrealtimeindicators/3august2023), the ONS publishes statistics more rapidly than traditional economic measures allow. They themselves say this about the real-time metrics: "These faster indicators are created using rapid response surveys, novel data sources and experimental methods." Translated into non-economist language, thi smeans that they're still not sure if these are reliable statistics, but they think so, and hope that they'll be useful.

I believe that I first heard about the Pret data on the Slate Money podcast (https://slate.com/podcasts/slate-money), but I'm not 100 percent sure. Regardless, I find this data to be quite interesting to observe and play with, because it is an attempt to grasp how many people are buying food, in person, at a variety of stores in different locations. The food and prices at each store are presumably identical, which means that the buying patterns are more reflective of where people are physically located, rather than what they want to eat or the price.

Oh, and just FYI: I’ve been to London many times, but haven’t ever eaten at a Pret a Manger. So I have no idea how good they are, but I have to assume that they’re doing something right, if they have survived for so many years, including through a pandemic.

Data and questions

This week's data is from the Office for National Statistics, and is one of the metrics they list in their real-time data project. You can download the data in Excel format from this page:

https://www.ons.gov.uk/economy/economicoutputandproductivity/output/datasets/transactionsatpretamanger

The excel file itself is at

/content/files/file.xlsx 

It's important to understand that the data set measures the number of in-person transactions at each store. It isn't looking at the amount paid per person, or the number of people eating each meal. It also ignores online and delivery sales, so we're really measuring in-person foot traffic at each store.

However, the data isn't broken down by store. Rather, each Pret location is put into one of several categories:

• Yorkshire
• London: Suburban
• Manchester
• London: West End
• London: Stations
• Regional Towns
• Scotland
• London: Airports
• London: City Worker
• Regional Stations

A full description of these locations, and of the overall methodology for this study, can be found at https://www.ons.gov.uk/economy/economicoutputandproductivity/output/methodologies/coronavirusandthelatestindicatorsfortheukeconomyandsocietymethodology .

In the first four weeks of 2020, the ONS and Pret a Manger calculated the average number of in-person transactions that took place in stores in each regional category. So they calculated the average number of weekly transactions for all stores in Yorkshire during those four weeks, then the average number of weekly transactions for all stores in suburban London during those four weeks, etc. That calculation was set to be the base, for a value of 100.

Starting in April 2021, the ONS calculated the average number of weekly transactions across stores in each regional category, and compared that number with the original base of 100. That's the number in the data set.

This means that numbers above 100 indicate an increase since those first four weeks in 2020, whereas numbers below 100 indicate a decrease. If the numbers at a regional category increases from week to week, then those shops are doing increasingly well -- and vice versa.

I have seven questions and tasks for you this week. Not only do the learning goals include working with time series, selecting rows and columns, choosing axes for methods, and resampling, but I also want to ask you not to use assignment after you create the data frame. Instead, we'll use method chaining for each of the questions – which means, if you need to create one or more new columns, you'll be using the `assign` method:

• Read the data into a data frame, treating the "Week Ending" column as dates and using it as the index.
• What three categories of stores are doing best, as of the latest data, vs. the baseline?