Not really same size k-means

[zamarajev]

Hi,

I am facing an issue, where the clusters are far different in size.
Please find my dataset and code enclosed.
For the initial centroid initialization, I am using kmeans++.
All of the requirements from requirements.txt are fulfilled.

Here is the output:
Cluster 1 : 13
Cluster 2 : 13
Cluster 3 : 10
Cluster 4 : 13
Cluster 5 : 12
Cluster 6 : 13
Cluster 7 : 13
Cluster 8 : 13
Cluster 9 : 0
Cluster 10 : 0
Total : 100

Is this a normal behavior or is it a bug?
For my case the cluster size is crucial, so I need the sizes to differ in max 1-2 objects.
Is this possible?
Can a more balanced sizing be achieved?

Thanks in advance!

dataset and code.zip

[zamarajev]

Sorry, I was setting a wrong amount of clusters in the constructor.
This is the output now:

Cluster 1 : 10
Cluster 2 : 10
Cluster 3 : 10
Cluster 4 : 10
Cluster 5 : 10
Cluster 6 : 10
Cluster 7 : 10
Cluster 8 : 10
Cluster 9 : 10
Cluster 10 : 10
Total : 100