Web scraping of food delivery applications data

The way people get food changed drastically during the last years, especially during the well-known pandemic of COVID-19 that occurred. People would go less to the grocery store and rely more extensively on online food delivery services like Deliveroo, Uber Eats and Takeaway. These websites produce a big amount of data which can be used to produce some interesting statistical studies. Collecting this data is a big challenge as these websites do not offer their data about their food and services to anyone because they are private, profit companies. A webscraping tool was therefore needed to collect all the information available on their websites. The only information available is the actual information about the restaurants and their menus, the rest is impossible to access. This work covers the elaboration and the resulting databases of the web scrapers that were built. An overview of some possible statistical analyses of the data is presented using geographical representations and price distributions in Flanders. Sciensano, the national public health agency at the origin of this project, will be able to use these data to conduct studies on the relation between the health of the Flemish citizens and the availability of food and their type on the different platforms. This work focuses on the different methods that were found to get the data from the three main food delivery platforms in Flanders with a goal to provide scripts that could be reused on different points of interest. Some interesting statistics can be computed from this kind of data and this work provides some basic statistical and geographical analyses of the price of food items per platform, per location and per restaurant. Studies of this kind have already been conducted but none of them were in Belgium and a thorough investigation of the websites structure was needed to be able to scrape all the required data.