As a simplified example, my data looks like this:
I would like to cluster my points with MLTK's DBSCAN, using radius
for the eps
parameter.
My radius
column is the result of an eval
function. All the rows have the same value.
I am trying to use this command:
… | fit DBSCAN eps=radius metric*
But I get this error:
Error in 'fit' command: Error while initializing algorithm "DBSCAN": Invalid value for eps: must be a float
Even though when I use isnum(radius)
the result is TRUE
Hi,
You need to specify eps=0.3 instead of eps=radius as eps accepts only numeric value
http://docs.splunk.com/Documentation/MLApp/2.3.0/User/Algorithms#DBSCAN
Let me know if this solves your issue.
Thanks,
Specifying eps=0.3 works.
However, my "radius" value is a result of previous data manipulation (I run several SPL commands beforehand to obtain that value). If I change my data source but run the same commands, "radius" will change. So for example, it could range from 0.1 to 1.0
This is why I would like to use a variable "radius," rather than a hard-coded number.