advatage: Compared to normally K-means algorithm, the value of k we set is not very necessary. The new algorithm will automatically find the value of which is close to the real one.
dataset: data02.txt
Adaptive K-means algorithm perfomances:
k==10:
k==8:
k==6:
k==4:
Normal K-means algorithm performance:
k==10:
k==8:
k==6:
k==4:
from the abrove performance, we can see that Adaptive K-mean is different from Normal K-means. Adaptive K-means can separate the points to different cluster well no matter what value k takes. And Normal K-means can not separate the points well because it is highly dependent on the k.