mpl: Maximum Adjusted Profile Likelihood Estimation - Generic...

In hoa: Higher Order Likelihood Inference

Description

Calculates the maximum adjusted profile likelihood estimates.

Usage

mpl(fitted, ...)

Arguments

fitted	any fitted model object for which the maximum adjusted profile likelihood estimates can be calculated.
...	absorbs any additional argument.

Details

This function is generic (see methods); method functions can be written to handle specific classes of data. Classes which already have methods for this function include: nlreg.

Value

the maximum adjusted profile likelihood estimates for all parameters of a regression model or for a subset of them.

See Also

mpl.nlreg, nlreg.object, methods

Examples

data(metsulfuron)
metsulfuron.nl <- 
    nlreg( formula = log(area) ~ log( b1+(b2-b1) / (1+(dose/b4)^b3) ), 
           weights = ~ ( 1+dose^exp(g) )^2, data = metsulfuron, hoa = TRUE, 
           start = c(b1 = 138, b2 = 2470, b3 = 2, b4 = 0.07, g = log(0.3)) )
mpl( metsulfuron.nl, trace = TRUE )
##
options( object.size = 10000000 )
data(chlorsulfuron)
chlorsulfuron.nl <- 
    nlreg( log(area) ~ log( b1+(b2-b1) / (1+(dose/b4)^b3) ), 
        weights = ~ ( 1+k*dose^g*(b2-b1)^2/(1+(dose/b4)^b3)^4*b3^2*dose^(2*b3-2)/
                    b4^(2*b3)/(b1+(b2-b1)/(1+(dose/b4)^b3))^2 ),
        start = c(b1 = 2.2, b2 = 1700, b3 = 2.8, b4 = 0.28, g = 2.7, k = 1), 
        data = chlorsulfuron, hoa = TRUE, trace = TRUE,  
        control = list(x.tol = 10^-12, rel.tol = 10^-12, step.min = 10^-12) )
mpl( chlorsulfuron.nl, trace = TRUE )

Example output

differentiating mean and variance function -- may take a while

iteration 1 : modified profile log likelihood = 0.3964742 
iteration 2 : modified profile log likelihood = 0.3964742 
Formula:
log(area) ~ log(b1 + (b2 - b1)/(1 + (dose/b4)^b3))
Variance function:
~(1 + dose^exp(g))^2 * exp(logs)

Higher order method used: Skovgaard's r*

Variance parameters
     MMPLE    MLE 
g    -1.274 -1.259
logs -3.716 -3.820

Regression coefficients
      MMPLE       MLE 
b1 1.389e+02 1.389e+02
b2 2.471e+03 2.471e+03
b3 1.709e+00 1.709e+00
b4 7.727e-02 7.727e-02

Total number of observations: 40
Total number of parameters: 6
-2*Log Lmp -0.7929 

Algorithm converged in 2 iterations

iteration 1 : log likelihood = -35.12366 
iteration 2 : log likelihood = -35.10621 
iteration 3 : log likelihood = -35.10528 
iteration 4 : log likelihood = -35.10515 
iteration 5 : log likelihood = -35.10513 
iteration 6 : log likelihood = -35.10513 
iteration 7 : log likelihood = -35.10513 
iteration 8 : log likelihood = -35.10513 
iteration 9 : log likelihood = -35.10513 
iteration 10 : log likelihood = -35.10513 

differentiating mean and variance function -- may take a while

iteration 1 : modified profile log likelihood = -35.92606 
iteration 2 : modified profile log likelihood = -35.92892 
iteration 3 : modified profile log likelihood = -35.92967 
iteration 4 : modified profile log likelihood = -35.92967 
Formula:
log(area) ~ log(b1 + (b2 - b1)/(1 + (dose/b4)^b3))
Variance function:
~(1 + k * dose^g * (b2 - b1)^2/(1 + (dose/b4)^b3)^4 * b3^2 * 
    dose^(2 * b3 - 2)/b4^(2 * b3)/(b1 + (b2 - b1)/(1 + (dose/b4)^b3))^2) * 
    exp(logs)

Higher order method used: Skovgaard's r*

Variance parameters
     MMPLE    MLE 
g     2.645  2.603
k     1.081  1.010
logs -1.825 -1.888

Regression coefficients
      MMPLE       MLE 
b1    2.2318    2.2063
b2 1657.3476 1657.3485
b3    2.8582    2.8426
b4    0.2752    0.2758

Total number of observations: 51
Total number of parameters: 7
-2*Log Lmp 71.86 

Algorithm converged in 4 iterations

hoa documentation built on May 2, 2019, 8:56 a.m.