“the catch is that the distribution of check-in densities in the real world varies greatly as a function of venue shape and size, which in turn are a function of category or chain, and location. So, the check-ins at a suburban Target look very different from the check-ins at a bar in Manhattan. There is no single value of MinPts and Eps that would work globally.”

To anyone reading this who wants to try DBSCAN, give its more recent brother HDBSCAN a go. The H stands for hierarchical and it's way better when dealing with clusters that aren't very similar (think big vs small).

There's an excellent pip package that seamlessly integrates woth scikit-learn too!