estimate phi

Share:

Description

Estimate phi in a flexible manner as guided by lframe - usually involving multiple strata, and allowing for formulas other than the default choice

Usage

1
phi.update.lframe(obj, return.fit = FALSE, lframe, glm.frm = NULL, ...)

Arguments

obj

a matrix whose rowSums are used to estimate phi

return.fit

logical - the result of the fit should be returned instead of phi

lframe

A data.frame with columns y, x, strata, orig (or matching those in glm.frm) and for which rownames(lframe)[lframe$orig] match the rownames of obj.

glm.frm

a formula. If omitted y ~ bs( x, knots= c(50,100) ) will be used.

...

curently not used

Details

fitting phi - the probabilities that a sonicant lands in a particular bin - is crucial to estimating theta, then number of sonicants when more than one lans in a bin. bins may be defined by sonicant lengths when there is just one sample. When there are multiple samples, then one may wish to estimate phi separately for each one. lframe$strata indicates separate sample and the default glm.frm will fit each one separately.

Value

a vector whose sum is 1.0 with one element for each row of obj

Author(s)

Charles C. Berry ccberry@users.r-forge.r-project.org

See Also

maxEM