biomod2: Find.Optim.Stat – R documentation

Become an expert in R — Interactive courses, Cheat Sheets, certificates and more!

Find.Optim.Stat

Calculate the best score according to a given evaluation method

Description

Find.Optim.Stat is an internal biomod2 function to find the threshold to convert continuous values into binary ones leading to the best score for a given evaluation metric.

Usage

Find.Optim.Stat(Stat='TSS',
                  Fit,
                  Obs,
                  Nb.thresh.test = 100,
                  Fixed.thresh = NULL)

Arguments

`Stat`	either 'ROC', TSS', 'KAPPA', 'ACCURACY', 'BIAS', 'POD', 'FAR', 'POFD', 'SR', 'CSI', 'ETS', 'HK', 'HSS', 'OR' or 'ORSS'
`Fit`	vector of fitted values (continuous)
`Obs`	vector of observed values (binary)
`Nb.thresh.test`	integer, the numer of thresholds tested over the range of fitted value
`Fixed.thresh`	integer, if not `NULL`, the only threshold value tested

Details

Please refer to BIOMOD_Modeling to get more information about this metrics. If you give a Fixed.thresh, no optimisation will be done. Only the score for this threshold will be returned.

Value

A 1 row x 4 column matrix :

best.iter: the best score obtained for chosen statistic
cutoff: the associated cut-off used for transform fitted vector into binary
sensibility: the sensibility with this threshold
specificity: the specificity with this threshold

Author(s)

Damien Georges

Examples

a <- sample(c(0,1),100, replace=TRUE)

  ##' random drawing
  b <- runif(100,min=0,max=1000)
  Find.Optim.Stat(Stat='TSS',
                  Fit=b,
                  Obs=a)

  ##' biased drawing
  BiasedDrawing <- function(x, m1=300, sd1=200, m2=700, sd2=200){
    return(ifelse(x<0.5, rnorm(1,m1,sd1), rnorm(1,m2,sd2)))
  }

  c <- sapply(a,BiasedDrawing)

  Find.Optim.Stat(Stat='TSS',
                  Fit=c,
                  Obs=a,
                  Nb.thresh.test = 100)

biomod2

Ensemble Platform for Species Distribution Modeling

v3.4.6

GPL-2

Authors

Wilfried Thuiller [aut, cre], Damien Georges [aut, cre], Robin Engler [aut], Frank Breiner [aut]

Initial release

2020-02-24