Suppose we have the vector
y = [1 1 1 2 3 4]
Then, with eps = 1 and k = 4 we clearly expect that 3 to be on the boundary of
the group formed by [1 1 1 2]. By algorithm definition, 3 is not allowed to
propagate group membership. However, we find that
>> [a,b] = dbscan(y', 3, 1)
a =
1 1 1 1 1 1
b =
1 1 1 1 0 0
where 3 did propagate the group membership to 4. 4 However should be classified
as an outlier. (Note that I entered k-1 to dbscan, as explained in my other bug
report.
Original issue reported on code.google.com by fre...@googlemail.com on 29 Nov 2012 at 3:35
Original issue reported on code.google.com by
fre...@googlemail.com
on 29 Nov 2012 at 3:35