IoT solutions, by definition if not implementation, can generate a lot of data. Indeed, most IoT solutions observe the world in one or often several ways. Those observations generate data at whatever rate the solution specifies (called a sample rate). To make the data most useful for historical or similar analytics, you have to save the data for later processing. Database systems provide a perfect solution for storing IoT data and making it available for later use.
With the exception of the last project, you wouldn’t consider storing the data from the example projects ...