bigrfprox-class: Proximity Matrices
Description
Class representing a proximity matrix for a random forest. Element [m, n] in the matrix indicate the number of times examples m and n end up in the same node, averaged over the number of trees in the forest. This is useful for computing scaling co-ordinates, prototypes, and outliers.
Objects from the Class
Objects can be created by calls of the form new("bigrfprox", ...), but most often are generated by proximities. "bigrfprox" inherits from "big.matrix". Given a "bigrfprox" object prox, each row represents an example, and each column represents the ncol(prox) "nearest" examples. If ncol(prox) is less than the number of examples, then the examples referred to in each row could be different. In that case, the big.matrix prox@examples indicates which examples are referred to in prox, i.e. prox[i, j] is the proximity between example i and the example prox@examples[i, j].Slots
examples:- Object of class
"big.matrix". If the number of columns in the proximity matrix is less than the number of rows, this indicates which examples are referred to in the proximity matrix. Otherwise (i.e. all pairwise proximities were computed), it is NULL. cachepath:- Object of class
"character". The cache path for the big.matrix's for the proximity matrix and examples, or NULL if the big.matrix's were created in memory. address:- Object of class
"externalptr". Inherited from class "big.matrix".
Extends
Class "big.matrix", directly.Methods
- scaling
signature(prox = "bigrfprox"): Compute metric scaling co-ordinates for training examples in the random forest. See scaling for details.- prototypes
signature(forest = "bigcforest", prox = "bigrfprox"): Compute class prototypes. See prototypes for details.