navigate: Limit which points are included in the nearest neighbor...

In mdlincoln/pathway: Find Point Pathways Through a Matrix

Description

When seeking the next point in a space, search only among those points that fulfill a given function based on the previously-selected points.

Usage

navigate(.p, ...)

navigate_ordered(x, pi, p1, p2, n)

navigate_ordered_desc(x, pi, p1, p2, n)

navigate_unique(x, pi, p1, p2, n)

navigate_any(x, pi, p1, p2, n)

Arguments

.p	A function that returns a vector of indices
...	Additional arguments passed on to .p
x	The original matrix
pi	Indices of already-selected-nodes. During the first step of a pathway, pi will be NULL
p1	Start and end points used in pathway
p2	Start and end points used in pathway
n	Number of indices to be found

Details

The supplied functions navigate_ordered, navigate_unique, and navigate_any can be supplied directly to pathway.

navigate is a generator that takes an arbitrary user-supplied predicate function that will generate a vector of search indices when given x, pi, p1, and p2. This can be useful for drawing a path that, for example, only moves forward in time, or which introduces a selection probability based on attributes of the point supplied in another table.

Value

Integer. An integer vector of indices within x forming the search space for the next step of the nearest neighbor search.

Functions

• navigate_ordered: Select only points with indices following the ones already selected.

• navigate_ordered_desc: Select only points with indices preceding the ones already selected.

• navigate_unique: Select only points that have not yet been visited.

• navigate_any: Select any point, even ones that have already been visited.

Examples

m <- matrix(runif(1000), nrow = 500, ncol = 2)
obs_types <- sample(c("setosa", "versicolor", "virginica"), 500, replace = TRUE)

# A custom predicate function must take the original matrix, the list of
# previously-selected pathway points, along with p1 and p2.
different_species <- function(x, pi, p1, p2, obs_types) {
  if (is.null(pi)) {
    search_space <- 1:nrow(x)
  } else {
    # Only search observations that do not have the same species as the immediately previous one.
    prev_type <- obs_types[tail(pi, 1)]
    search_space <- which(obs_types != prev_type)
  }

  # Don't forget to exclude p1 and p2
  setdiff(search_space, c(p1, p2))
}

p_species <- pathway(m, 2, 11, n = 8, navigator = navigate(different_species, obs_types))

mdlincoln/pathway documentation built on May 23, 2019, 1:13 p.m.