css: Optimal sample size for infrared spectroscopic data

In AlexandreWadoux/soilspec: The soilspec R package

Description

Function to determine the optimal number of spectra to be sent to the laboratory for soil analysis. This function works by comparing the probability density function (pdf) of the population and the pdf of the sample set in order to asses the representativeness of the sample. The two pdfs are compared based on the mean euclidean distance (msd). See Ramirez-Lopez, et al. (2014) for more details.

Usage

css(S, k, method = "kms", repetitions = 10, n = 512, from, to, bw, ...)

Arguments

S	A matrix of the scores of the principal components.
k	A vector containing the sample set sizes to be evaluated.
method	The sampling algorithm. Options are: (i) kss (Kennard-Stone sampling), (ii) kms (K-means sampling) the default, (iii) clhs (conditioned Latin Hypercube sampling).
repetitions	The number of times that the sampling must be carried out for each sample size to be evaluated. The results of the final msd is the average of the ones obtained at each iteration. Note that since the kss method is deterministic and always return the same results, there is no need for repetitions.
n	The number of equally spaced points at which the probability densities are to be estimated (see density).
from, to	A vector of the left and right-most points of the grid at which the densities are to be estimated. Default is the minimums and maximums of the variables in S.
bw	A vector containing the the smoothing bandwidth to be use for the probability densities (see density).
...	Arguments to be passed to the calibration sampling algorithms, i.e. additional aruments to be used for the clhs, kenStone or naes functions which run inside this function.

Value

A table with the following columns:

css	The sample size (k).
msd	Value of the msd for each sample size.
msd_sd	The standard deviation of the msd for all the repetitions (does not apply to kss since it always return the same results).

References

Ramirez-Lopez, L., Schmidt, K., Behrens, T., van Wesemael, B., Dematte, J. A., Scholten, T. (2014). Sampling optimal calibration sets in soil infrared spectroscopy. Geoderma, 226, 140-150.

AlexandreWadoux/soilspec documentation built on March 7, 2021, 1:54 p.m.