Fucang Jia
2004-Mar-09 15:29 UTC
[R] How to ascertain the number of clusters automatically?
Hi, everyone, There is many small cells which can be classified into several big cells from the scanned image. K-means clustering does not work well in this condition. I have done hierarchical clustering on cells successfully which uses shortest distance between classes. The number of clusters is about 3, 4, 5, 6, 7 generally. One can ascertain the number of clusters visually. But because there are thousands of images to be clustered. So it is humdrum to me. I want to know if there are any methods that can be used to ascertain the number of clusters automatically, especially in this case, only several clusters? Thank you very much! Best, Fucang
Hanke, Alex
2004-Mar-10 13:05 UTC
[R] How to ascertain the number of clusters automatically?
Hi, You may be interested in a clustering algorithm called OPTICS. It is both interactive and automatic and does not require a lot of input parameters. It is described as creating " an augmented ordering of the data representing its density-based clustering structure". It "automatically and efficiently extracts not only traditional clustering information but also the intrinsic clustering structure". Check out this site: http://www.dbs.informatik.uni-muenchen.de/Forschung/KDD/Clustering/ Alex -----Original Message----- From: Fucang Jia [mailto:jiafucang at hotmail.com] Sent: March 9, 2004 11:30 AM To: r-help at stat.math.ethz.ch Subject: [R] How to ascertain the number of clusters automatically? Hi, everyone, There is many small cells which can be classified into several big cells from the scanned image. K-means clustering does not work well in this condition. I have done hierarchical clustering on cells successfully which uses shortest distance between classes. The number of clusters is about 3, 4, 5, 6, 7 generally. One can ascertain the number of clusters visually. But because there are thousands of images to be clustered. So it is humdrum to me. I want to know if there are any methods that can be used to ascertain the number of clusters automatically, especially in this case, only several clusters? Thank you very much! Best, Fucang ______________________________________________ R-help at stat.math.ethz.ch mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html