1
$\begingroup$

I have some 1 dimensional data. Each record in the data is a specific time of the day. In order to cluster it I projected the data onto a circle of radius 1 unit. Now I need to find clusters in this data. The number of clusters are unknown and it is preferred to find clusters with high density of records in them. By density I mean that a large volume of records should be packed in a small space.

How should I go about finding clusters in the above mentioned data?

$\endgroup$
1
$\begingroup$

Instead of projecting into the circle and thus making your problem 2d, why don't you just use a cyclic distance measure?

This problem should be straightforward by doing kernel density estimation on the (cyclic) time of day. Then find the peaks, which are your clusters.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.