search

muesli / kmeans

k-means clustering algorithm implementation written in Go
MIT License
457 stars 51 forks source link

Feature Request: Allow observations with gaps #13

Open guenhter opened 4 years ago

guenhter commented 4 years ago

Hi,

do you mind if we add the option to your kmeans to ignore gaps in observations.

The main issue we have is: We have observations with a few hundred dimensions. Unfortunately, some of the observations have gaps but we still want to consider those observations. Currently, we just fill gaps with 0 what is actually not the right way because it falsifies the result. What we actually would like to have is, to ignore gaps, so we don't consider the particular dimension for that particular observation at all.

Does this make sense to you?

muesli commented 4 years ago

Yes, I think that sounds reasonable.