mfm_sampleSize: Choice of the optimal sample size for the MFM algorithm

In AlexisDerumigny/MFunctMatching: Subsample Selection by Multifunctional Matching

Description

Choice of the optimal sample size for the MFM algorithm

Usage

mfm_sampleSize(
  data,
  rangeSizeSample,
  alpha = 0.9,
  listfns = NULL,
  listinputs = NULL,
  coordX = NULL,
  coordY = NULL,
  minDistance = 0,
  nDraws = 2000,
  nReplications = 1000,
  methodNormalization = "ecdf",
  weightsfns = NULL,
  progressBar = TRUE
)

Arguments

data	the dataframe to use
rangeSizeSample	the list of the sample sizes to tested
alpha	the percentage of reduction of the criterion
listfns	list of functionals to be used. If NULL (the default) then the mean, the standard deviation and Kendall's tau between each variables are used.
listinputs	list of inputs to the functionals. Only used when listfns is not NULL.
coordX	X coordinates of the points. Used for spatial data.
coordY	Y coordinates of the points. Used for spatial data.
minDistance	minimum distance between two observations of the selected sample
nDraws	number of draws to compute each MFM criterion.
nReplications	number of replications of the MFM criterion computation.
methodNormalization	the method used for normalization. Can be ecdf for normalization by the empirical cumulative distribution function or meansd for the normalization (X-E[X])/sd(X).
weightsfns	weights for each of the functionals. These weights can be used to give more or less importance to some of the functionals. It should be a vector of the same length as the listfns.
progressBar	TRUE: display a progressbar

Examples

variables = c("mass", "moist", "elev")
df = agridat::gartner.corn
result = mfm_sampleSize(
  data = df[,variables], rangeSizeSample = c(15,25,30),
  nDraws = 500, nReplications = 20,
  coordX = df$long, coordY = df$lat)

AlexisDerumigny/MFunctMatching documentation built on Dec. 31, 2020, 9:47 a.m.