Introduction to k-means Clustering

Hopefully, by now, you can see that finding clusters is extremely valuable in a machine learning workflow. However, how can you actually find these clusters? One of the most basic yet popular approaches is by using a cluster analysis called k-means clustering. k-means works by searching for K clusters in your data and the workflow is actually quite intuitive – we will start with the no-math introduction to k-means, followed by an implementation in Python.

No-Math k-means Walkthrough

Here is the no-math algorithm of k-means clustering:

1. Pick K centroids (K = expected distinct # of clusters).
2. Randomly place K centroids anywhere amongst your existing training data.
3. Calculate the Euclidean distance from each centroid to all the points in your training data.
4. Training data points get grouped in with their nearest centroid.
5. Amongst the data points grouped into each centroid, calculate the mean data point and move your...