Sei sulla pagina 1di 1

Summary of Coursera Process Mining.

Week 1 Lecture 7: Cluster Analysis


Created by Ariadna73 using a beta version of EASY+

(Please click, and share the video. Thank you, Ari!)

This is another unsupervised technique


k-means technique
Basic idea
We have instances with attributes
We want to "cluster" them into homogeneous groups
How does it work?
First, you set up your "Centroids"
- You can do it arbitrarily
- But it is better to do it in a smarter way!
Then you analyze each instance and assign it to the nearest centroid
Then you calculate the average distance of every instance to its centroid and create a new one
Then you analyze the instances again, recalculate the centreoids and repeat until there are no
changes
Other clustering techniques
Agglomerative hierarchical clustering
Put the instances in different hierarchies and draw horizontal lines to create the clusters

The main idea is that the instances in one cluster are similar to each other and very
different from the other clusters
Then, with those homogeneous sets, you can more easily apply other data mining methods we are
studying
The teacher then describes an example with clients in a restaurant
He shows how the results look in Rapid Miner

Clustering can be used to split and analyze event logs for process mining

____________________________________________________________________________________
Created by Ariadna73 using a beta version of EASY+ (Please click, and share the video. Thank you, Ari!)
Page 1 of 1

Potrebbero piacerti anche