Number of Categories for K-Means: Elbow and Silhouette Methods
K-Means is a very common and powerful clusterization algorithm widely used in an unsupervised machine learning tasks for dividing data into categories. The only decision you have to make is the number of clusters you want your data to be divided into — k number.
Sometimes you already know how many categories you need to have. It depends a lot on the type of your problem, your data, and the problems you are solving.
Categorizing Instagram Tags with K-Means
Over the last couple of years Instagram, Facebook and many other social media have gotten rid of the chronological order in their post feed. While frustrating at first, this decision encouraged one part of social media that I like most of all: your content can be seen, discovered and rated not only by your friends and followers, but also by many other new people. To make your content discoverable you can use features such as hashtags, geolocations, tagging other people and so on.