search

khunreus / cluster-categorical

14 stars 6 forks source link

any other way to cluster ordinary data？ #1

Open Hui2019-creat opened 2 years ago

Hui2019-creat commented 2 years ago

Hi Anastasia Reusova, Recently, I am clustering some ordinary data (data such as '0,1,2,3', the bigger the worse). I tried the cluster method of Kmodes in R and the way you suggested ( cluster-categorical). However, both of them output poor results. Is there any other way that I can use? Or it is due to my poor original data (200 subjects and 8 variables)

AndreM84 commented 2 years ago

Hi @Hui2019-creat,

there are plenty of methods which you could try to use after getting an understanding of their individual 'principles' (which means distance meausres, standardadization methods, clustering algorithms).

While the blog is focusing on Hierarchical Clustering you could e.g. try also Partitioning Around Medoids (PAM) or the k-modes clustering apporach.

The package NbClust seems to allow an easy adaption of the k-means and hierarchical clustering approaches with regards to distance measures and methods for assigning objects to clusters.

A few minutes ago I found the EnsCat R package, which seems to be another alternative, although I did not dive into details yet.

best regards André

Hui2019-creat commented 2 years ago

谢谢 已收到