leia: Integrate AUC for curves from a multigroup GAM
In wfforrest/maeve: Modeling Accessories Enabling Visualization of Experiments

Description Usage Arguments Value Author(s) References Examples

Integrate AUC for curves from a multigroup GAM

leia(
  model,
  N = 25,
  xmin = NULL,
  xmax = NULL,
  xname = NULL,
  xrange_norm_method = c("none", "xrange", "half_xrange_squared"),
  subtract_starting_value = TRUE,
  groupname = NULL,
  contrast = c("Identity", "Dunnett", "Tukey", "Sequen", "custom"),
  custom_contrast = NULL,
  reference_Dunnett = maeve_options("reference_Dunnett"),
  glht_rhs = NULL,
  glht_alternative = c("two.sided", "less", "greater"),
  provide_warnings = TRUE,
  single_group_identifier = "group_01",
  derivative = 0,
  return_list = FALSE
)

`model`	object of class 'gam' from package 'mgcv' (not package 'gam').
`N`	integer number of equally-spaced points to use in numerical integration.
`xmin`	numeric lower bound of definite integral
`xmax`	numeric upper bound of definite integral
`xname`	character name of numeric variable used in the gam smooth. If provided, it should be accessible in "model$model[[xname]]".
`xrange_norm_method`	character string determining how the xrange normalization within leia() should be done. Must be one of "none", "xrange", "half_xrange_squared".
`subtract_starting_value`	logical: Subtract from all computed AUCs the spline fit at xmin for that group (adjusted for xrange_norm_method choice).
`groupname`	character name of grouping factor by which the one-dimensional smooth is conditioned. If provided, it should be the first and only component in model$xlevel.
`contrast`	character string with contrast type. 'Identity' gives each curve separately. 'Dunnett', 'Tukey', etc. from multcomp::contrMat() are accepted, as is 'custom' to indicate a custom contrast.
`custom_contrast`	numeric matrix of contrasts for slope estimates. Column names must match exactly with treatment group names. If it is not NULL, it overrides whatever is specified in "contrast".
`reference_Dunnett`	character string with reference group for Dunnett's test. If provided, must match exactly to one of "model$xlevel[['group']]".
`glht_rhs`	numeric vector for the right-hand side (rhs) of the contrast hypothesis; passed to glht() "rhs = " argument.
`glht_alternative`	character string passed to glht() as its "alternative = " argument.
`provide_warnings`	logical: provide warning messages when something looks look awry?
`single_group_identifier`	character string with name to give a single group in the output.
`derivative`	integer specifying which derivative of the fixed smooth to integrate (the default, "0", means to just integrate the fixed smooth).
`return_list`	logical: return a whole bunch of output in a list.

an object of class glht from package multcomp (or a list if return_list == TRUE)

Bill Forrest <forrest@gene.com>

Bill Forrest forrest@gene.com

www.r-project.org

N <- 100
x <- seq( 0, pi, length = N )
set.seed( 20180425 )
dat <- data.frame( x, y = sin( x ) + rnorm( N, 0, .01 ) )
model <- mgcv::gam( y ~ s( x ), data = dat )
leia_out  <- leia( model, xmin = 0, xmax = pi/2, glht_rhs = 1 ) # about "1.0" over [ 0, pi/2 ]...
leia_list <- leia( model, xmin = 0, xmax = pi/2, glht_rhs = 1, return_list = TRUE ) # extras...