Reputation: 41
I am trying to cluster a dataset which has 25 columns - 1 column with an id and 24 columns telling how many times an application with this id has been clicked for each hour of the day. When I try to run dbscan from the package dbscan the R session is aborted with a fatal error.
I tried to use dbscan on the same data, where it was group by day instead of hour and it works fine. Is there a limit on how many dimensions dbscan can handle?
Upvotes: 3
Views: 755
Reputation: 3075
This was indeed an early bug which has been resolved in the current version. Here is some random data with 1000 points and 100 dimensions.
> library(dbscan)
> dat <- matrix(runif(100*1000), ncol=100)
> system.time(db <- dbscan(dat, eps = 3.2))
user system elapsed
0.272 0.000 0.264
> db
DBSCAN clustering for 1000 objects.
Parameters: eps = 3.2, minPts = 5
The clustering contains 2 cluster(s).
Available fields: cluster, eps, minPts
Upvotes: 0