Solved – How to interpret adjusted rand index result

clusteringmachine learning

ARI takes values between -1 and 1. When I check the clustering result i got ARI = 0.21. What does it mean?. What is good ARI value?

can you recommend a good resource on ARI?

Thanks!

Best Answer

Depends on your data set.

But in general, 0.20 is really bad.

Just remember that 0 is putting everything in random buckets.

So 0.20 is a result quite close to randomness.

And don't forget that these indexes often cannot be applied on real data. If you have labels, don't use clustering. For ARI you need to know the ground truth. But then you are probably comparing result A with result B for benchmarking, and then you probably are more interested in whether A or B is "better" as in 'more similar'.

Related Question