Cluster Analysis - Introduction to Partitional Clustering with an example
Key moments in this video
00:13 Partitional Clustering – introduction
01:05 Partitional Clustering – the general approach
02:32 Example with eight datapoints and choosing two points as initial means
05:32 Example showing the second iteration
07:37 Example showing the third iteration
Introduction to Partitional Clustering with an example
Partitional clustering algorithms create an initial partitioning of K clusters by choosing the K Means or centers and then use an iterative relocation technique that attempts to improve the partitioning by moving objects from one group to another based on an objective function used to assess the partitioning quality. The small example shown here illustrates, in a very simple manner, the working of a partitional clustering algorithm.
The steps are:
1. Compute distances of all points with the means.
2. Assign the points to the closest means.
3. From the clusters so obtained, recompute the means.
4. Repeat steps 1 to 3 until the stopping criteria is reached, for example, the means do not change.
Link to Data handling and Distance computation measures
• Data Handling and Proximity Measures