boxcox_itr_bias_cor {lidaRtRee} | R Documentation |
Inverse Box-Cox transformation with bias correction
Description
Inverse Box-Cox transform with bias correction as suggested by Pu & Tiefelsdorf (2015). Here 'varmod' is not the local prediction variance as suggested in the paper but the model residuals variance. For variance computation, uses 'n-p' instead of 'n-1', with 'p' the number of variables in the model.
Usage
boxcox_itr_bias_cor(x, lambda, varmod)
Arguments
x |
vector or raster values to be transformed |
lambda |
numeric. parameter of Box-Cox transformation |
varmod |
numeric. model residuals variance |
Value
a vector or raster
References
Xiaojun Pu and Michael Tiefelsdorf, 2015. A variance-stabilizing transformation to mitigate biased variogram estimation in heterogeneous surfaces with clustered samples. doi:10.1007/978-3-319-22786-3_24
See Also
boxcox_tr
Box-Cox transformation,
boxcox_itr
inverse Box-Cox transformation.
Examples
x <- 1:10
boxcox_itr(x, 0.3)
boxcox_itr_bias_cor(x, 0.3, 0)
boxcox_itr_bias_cor(x, 0.3, 2)
# plot functions
curve(boxcox_itr(x, 0.3), 0, 3,
col = "blue",
main = "inverse Box Cox transf., lambda = 0.3",
xlab = "x", ylab = "inverse Boxcox(x, lambda = 0.3)"
)
curve(boxcox_itr_bias_cor(x, 0.3, 1), 0, 3, col = "red", add = TRUE)
curve(boxcox_itr_bias_cor(x, 0.3, 2), 0, 3, col = "black", add = TRUE)
legend("topleft",
legend = c(
"residuals variance = 2",
"residuals variance = 1", "residuals variance not accounted for"
),
col = c("black", "red", "blue"), lty = 1
)
[Package lidaRtRee version 4.0.5 Index]