Introduction to DBSCAN

As mentioned in the previous section, the strength of DBSCAN becomes apparent when we analyze the benefits of taking a density-based approach to clustering. DBSCAN evaluates density as a combination of neighborhood radius and minimum points found in a neighborhood deemed a cluster.

This concept can be driven home if we re-consider the scenario where you are tasked with organizing an unlabeled shipment of wine for your store. In the past example, it was made clear that we can find similar wines based off their features, such as scientific chemical traits. Knowing this information, we can more easily group together similar wines and efficiently have our products organized for sale in no time. Hopefully, that is clear by now – but what may not have been clear is the fact that products that you order to stock your store often reflect real-world purchase patterns. To promote variety in your inventory, but still have enough stock of the most popular wines, there is a highly...