Add labels for each data point in a plot

Question

I am having some issues with plotting my clusters in R. I am using Kmeans as a way of compressing my dataset. My dataset is already labeled into 10 classes, so i already know what class each datapoint belongs to. Is it possible to visualize how each datapoint is sorted within each cluster. Such that I could make some form of conclusion of the class distribution within each cluster.

I tried using table

table(data$labels[kmeans_output$clusterr==1])

which outputs

C0  C1  C2  C3  C4  C5  C6  C7  C8  C9 
  4   0   0   1   0  13   0   0 245   0

Which i think is the class distribution of cluster 1.

Is possible to visualize this for all the clusters?

That does not look like clustering worked well. – Has QUIT--Anony-Mousse Apr 05 '16 at 09:41 — Has QUIT--Anony-Mousse, Apr 05 '16 at 09:41

score 1 · Accepted Answer · edited May 23 '17 at 12:07

1

An example using iris as your example is not reproducible. Is this what you're looking for?

km <- kmeans(iris[, 1:2], 3)
plot(iris[, 1:2], col=km$cluster)
table(iris$Species, km$cluster)

edited May 23 '17 at 12:07

Community

1
1

answered Apr 05 '16 at 08:20

Vincent Bonhomme

7,235
2
27
38

how would you plot it, i am using 30 clusters or more so plotting it would be a bit dense. bet yes table works. – Lamda Apr 05 '16 at 08:28
well, any plotting approach on the same graph would likely superimpose all informations.. – Vincent Bonhomme Apr 05 '16 at 09:07
1

maybe a confusion matrix would be a better display? – Vincent Bonhomme Apr 05 '16 at 09:11
the table is different sized.. so ordinary confusiomatrix won't work – Lamda Apr 05 '16 at 10:10
you mean some levels are missing? – Vincent Bonhomme Apr 05 '16 at 10:16
i think `image.plot` seems to fit.. but labels aren't accurate – Lamda Apr 05 '16 at 11:06

Add labels for each data point in a plot

1 Answers1