sephora (version 0.1.31)

phenopar_polygon: Phenological parameters estimation in mass

Description

Estimation of phenological parameters from a set of numeric vectors stored in a RData file. Output is saved as a RData file at the destination specified by dirToSave

Usage

phenopar_polygon(
  path = NULL,
  product = c("MOD13Q1", "independent"),
  data,
  frequency = 23,
  method = c("OLS", "WLS"),
  sigma = NULL,
  numFreq,
  delta = 0,
  distance,
  samples,
  basis,
  corr = NULL,
  k,
  trace = FALSE,
  numCores = 20,
  dirToSave,
  reportFileBaseName = "phenopar_progress",
  outputFileBaseName = "polygon"
)

Value

At the location specified by dirToSave, a file containing a matrix with nrow equal to the number of numeric vectors analyzed and 6 columns, is saved. The name of this file is:

paste0(tools::file_path_sans_ext(basename(path)), "_phenoparams.RData").

Arguments

path

character with full path of RData file containing numeric vectors to analyze.

product

character specifying whether dataset is the MOD13Q1 product (default) or a different one (independent).

data

matrix with dataset to analyze. Pertinent when product="independent" only.

frequency

integer giving number of observations per season. Default, 23.

method

character. Should OLS or WLS be used for smoothing each numeric vector in RData file specified in path?

sigma

numeric vector of length equal to frequency. Each entry gives the standard deviation of observations acquired at same day of the year. Pertinent when method=WLS.

numFreq

integer specifying number of frequencies to use in harmonic regression model.

delta

numeric. Default, 0. When regression problem is ill-posed, this parameter allows a simple regularization.

distance

character indicating what distance to use in hierarchical clustering. All distances in tsclust are allowed.

samples

integer with number of samples to draw from smoothed version of numeric vector to analyze. Used exclusively in Functional Principal Components Analysis (FPCA)-based regression.

basis

list giving numeric basis used in FPCA-based regression. See details.

corr

Default NULL. Object defining correlation structure, can be numeric vector, matrix or function.

k

integer, number of principal components used in FPCA-based regression.

trace

logical. If TRUE, progress on the hierarchical clustering is printed on console. Default, FALSE.

numCores

integer. How many processing cores can be used?

dirToSave

character. In which directory to save analysis results?

reportFileBaseName

character. What base name should be given to a progress report file? Default, phenopar_progress.

outputFileBaseName

character. What base name should be given to the output file? Default, polygon.

See Also

phenopar, getSpiralPlot, tsclust.

Examples

# \donttest{
dirOUTPUT <- system.file("data", package = "sephora")
BASIS <- drbasis(n=100, q=2)

polygon_deciduous <- deciduous_polygon
for(i in 1:nrow(polygon_deciduous)){
  polygon_deciduous[i,] <- vecFromData(data=deciduous_polygon, numRow=i)$vec
}

# --- In the following example 'numCores=2' for CRAN
# --- testing purposes only. In a real life example
# --- users are encouraged to set 'numCores' to a number
# --- that reflects the size of their data set as well
# --- as the number of available cores

phenopar_polygon(data=polygon_deciduous,
                 product="independent",
                 numFreq = 3, distance = "dtw2",
                 samples=100, basis=BASIS,
                 k=3, numCores=2,
                 dirToSave=dirOUTPUT,
                 outputFileBaseName = "deciduous")
                 
# --- Auxiliary function to read phenopar_polygon output,
# --- used below to define deciduous_params object                   
LoadToEnvironment <- function(RData, env = new.env()){
                              load(RData, env)
                              return(env)}
                 
# --- colors used in spiralPlot below
cgu <- rgb(173/255,221/255,142/255)
csos <- rgb(120/255,198/255,121/255)
cmat <- rgb(49/255, 163/255,84/255)
csen <- rgb(217/255, 95/255, 14/255)
ceos <- rgb(254/255, 153/255, 41/255)
cdor <- rgb(208/255, 209/255, 230/255)

colores <- c(cgu,csos,cmat,csen,ceos,cdor)

# --- how to get a SpiralPlot
listRDatas <- list.files(path=dirOUTPUT,
                         pattern=".RData",
                         full.names=TRUE)
                         
deciduous_params <- LoadToEnvironment(listRDatas[1])

getSpiralPlot(MAT=deciduous_params$output, 
              LABELS=month.name,
              vp_param=list(width=0.5, height=0.7))
vcd::grid_legend(x=1.215, y=0.125, pch=18, col=colores,
                frame=FALSE,
                labels=c("GU","SoS","Mat","Sen","EoS","Dor"),
                title="Params")
            
# --- cleaning up after work
unlink(paste0(dirOUTPUT, "/deciduous_phenoParams.RData"))
unlink(paste0(dirOUTPUT, "/phenopar_progress.txt"))# }