0.25
0.5
0.75
1.25
1.5
1.75
2
Stability for selecting the number of clusters: literature review, questions, and ideas
Published on Jul 28, 20076348 Views
Related categories
Chapter list
Clustering Stability — a literature review, <br>many questions, and a few ideas for answers00:00
Overview00:22
The principle of stability00:46
Stability as a tool for model selection in clustering02:00
Stabilty – the general principle08:00
The toy figure in favor of stability09:36
Generating artificial data sets10:45
Generating artificial data sets (2)12:23
How to use the clustering algorithm18:01
Distances between the clusterings18:18
Distances between the clusterings (2)18:53
Distances between the clusterings (3)19:37
Distances between the clusterings (4)20:14
Which clusterings to compare?20:55
Stability scores22:37
Normalization23:37
Normalization (2)25:24
Selecting K, finally26:00
Selecting K, finally (2)27:25
Stability in theory30:34
Negative results on stability31:13
Negative results on stability (2)32:47
Negative results on stability (3)33:29
First catch: large vs. small sample size37:21
Possible solution: “stability window”38:12
Second catch: attaining the global minimum39:31
Possible solution: exploring objective function40:31
Possible solution: exploring objective function (2)40:39
Possible solution: exploring objective function (3)41:53
Catch 3: What is “the right K”,<br> actually?47:37
The “correct K”, first approach47:59
The “correct” K, second approach49:40
Idea: hierarchy of cluster core sets54:35
Summary01:03:19