Hello, i have project in clustering using K-Means,
I have question, how to determine clusters in K-Means, because so far, scientist just create a clusters as ex : cluster_1, cluster_2, cluster_3, cluster_n
How if determine clusters in other as ex : cluster_germany, cluster_us, cluster_british etc ?
The centroid of each data labeled group correspond to the cluster prototipe
Mr Mohammadreza Zandehshahvar
: Do you have references sir?Mr Erik Cuevas : Do you can tell me detail sir?
With the K-Means algorithm you define the number of clusters, K, and therefore, you should have some prior knowledge of what cluster groups you want to break the data into. If you have no idea of the number, you run K-Means with different K settings, then plot to see which covers the groups of data best, or use another algorithm that determine number of clusters. Then use your review of the data to determine number of labels. Because you may want more clusters to break out Germany cluster into 2 clusters: Germany & Austria in your example.
Ryan Putranda Kristianto . What you are asking is interpreting what each cluster means, and it is quite subjective. Erik Cuevas has correctly mentioned using centroids to find out cluster affinity. You can also look into things like amount of variability in each cluster, range of values etc. In general, look for all types of descriptive measures and start comparing them cluster by cluster.
- Badges
- Science method