Nearest Neighbors

Description

Computes the nearest neighbors of descriptors in an FPset or APset object for use with the jarvisPatrick clustering function. Only one of numNbrs or cutoff should be given, cutoff will take precedence if both are given. If numNbrs is given, then that many neighbors will be returned for each item in the set. If cutoff is given, then, for each item X, every neighbor that has a similarity value greater than or equal to the cutoff will be returned in the neighbor list for X.

Usage

1
   nearestNeighbors(x, numNbrs = NULL, cutoff = NULL, ...)

Arguments

x

Either an FPset or an APset.

numNbrs

Number of neighbors to find for each item. If not enough neighbors can be found the matrix will be padded with NA.

cutoff

The minimum similarity value an item must have to another item in order to be included in that items neighbor list. This parameter takes precedence over numNbrs. This parameter allows to obtain tighter clustering results.

...

These parameters will be passed into the distance function used, either cmp.similarity or fpSim, for APset and FPset, respectively.

Value

The return value is a list with the following components:

indexes

index values of nearest neighbors, for each item. If cutoff is used, this will be a list of lists, otherwise it will be a matrix

names

The names of each item in the set, as returned by cid

similarities

The similarity values of each neighbor to the item for that row. This will also be either a list of lists or a matrix, depending on whether or not cutoff was used. Each similarity values corresponds to the id number in the same position in the indexes entry

Author(s)

Kevin Horan

See Also

jarvisPatrick trimNeighbors

Examples

1
2
3
4
   data(sdfsample)
   ap = sdf2ap(sdfsample)
   nnm = nearestNeighbors(ap,cutoff=0.5)
   clustering = jarvisPatrick(nnm,k=2,mode="a1b")

Want to suggest features or report bugs for rdrr.io? Use the GitHub issue tracker.