d2s: Dense to sparse conversion

In GMKMcharlie: Unsupervised Gaussian Mixture and Minkowski and Spherical K-Means with Constraints

Description

Convert data from dense representation (matrix) to sparse representation (list of data frames).

Usage

d2s(
  X,
  zero = 0,
  threshold = 1e-16,
  verbose= TRUE
  )

Arguments

X	A d x N numeric matrix where N is the number of data points — each column is an observation, and d is the dimensionality. Column-observation representation promotes cache locality.
zero	A numeric value. Elements in X satisfying abs(X[i] - zero) <= threshold are treated as zeros. Default 0.
threshold	A numeric value, explained above.
verbose	A boolean value. TRUE prints progress.

Value

A list of size N. Value[[i]] is a 2-column data frame. The 1st column is a sorted integer vector of the indexes of nonzero dimensions. Values in these dimensions are stored in the 2nd column as a numeric vector.

Examples

N = 2000L
d = 3000L
X = matrix(rnorm(N * d) + 2, nrow = d)
# Fill many zeros in X:
X = apply(X, 2, function(x) {
  x[sort(sample(d, d * runif(1, 0.95, 0.99)))] = 0; x})
# Get the sparse version of X.
sparseX = GMKMcharlie::d2s(X)
str(sparseX[1:5])

GMKMcharlie documentation built on May 29, 2021, 9:08 a.m.