kdist_info: k-dist graph calculations

In unmnn/evoxploit: Augment Evolution Features to a Longitudinal Study Dataset

Description

Estimates an appropriate value for the DBSCAN parameter ε based on the k-dist graph, as suggested in \insertCiteEster:DBSCAN1996evoxploit.

Usage

kdist_info(dist_obj, k = round(log(attr(dist_obj, "Size"))))

Arguments

dist_obj	A distance matrix of type dist.
k	Used for k-dist graph. If not set, it defaults to log(n), where n is the number of objects, as suggested in \insertCiteKailing:RIS2003evoxploit.

Value

A list containing the following elements:

• knn_dist_sorted: A double vector of sorted k-dist values. Element names represent indices of dist_obj.

• dist_to_line: A double vector of distances between knn_dist_sorted and the line between its first and last element.

• idx_opt: The index of the maximum value in dist_to_line.

• eps_opt: The idx_optth value in knn_dist_sorted.

• minPts: k.

References

\insertAllCited

unmnn/evoxploit documentation built on Oct. 28, 2020, 12:24 p.m.