K-means clustering
K-means is the most popular of the clustering techniques because of its ease of use and implementation. It also has a partner by the name of K-medoid. These partitioning methods create level-one partitioning of the dataset. Let's discuss K-means in detail.
K-means algorithm
K-means start with a prototype. It takes centroids of data points from the dataset. This technique is used for the objects lying in the n-dimensional space.
The technique involves choosing the K number of centroids. This K is specified by the user and is chosen considering various factors. It defines how many clusters we want. So, choosing a higher or lower than the required K can lead to undesired results.
Now going forward, each point is assigned to its nearest centroid. As many points get associated with a specific centroid, a cluster is formed. The centroid can get updated depending on the points that are part of the current cluster.
This process is done repeatedly until the centroid gets constant.
