iterations: Harvest Iteration Statistics from NONMEM Output and Convert...

In metrumrg: Pharmacometric Tools for Data Preparation, Modeling, Simulation, and Reporting

Description

iterations excises the relevant portion of the output file, converting the text to parameter and gradient estimates.

Usage

iterations(x, ...)

Arguments

x	character
...	ignored

Details

The default result has intermediate dimensions as a compromise between very wide and very tall.

Value

data frame indicating parameter estimates and gradients by iteration

Author(s)

Tim Bergsma

References

http://metrumrg.googlecode.com

See Also

• cov2cor

Examples

lst <- c(
'0ERROR SUBROUTINE CALLED WITH EVERY EVENT RECORD.',
'1',
' ',
' ',
' #METH: First Order Conditional Estimation with Interaction',
' MONITORING OF SEARCH:',
'',
'0ITERATION NO.:    0    OBJECTIVE VALUE:   3140.31595361523        NO. OF FUNC. EVALS.:10',
' CUMULATIVE NO. OF FUNC. EVALS.:       10',
' PARAMETER:  1.0E-1  1.0E-1  1.0E-1  1.0E-1  1.0E-1  1.0E-1  1.0E-1  1.0E-1  1.0E-1  1.0E-1',
'             1.0E-1',
' GRADIENT:   3.9E+0 -4.0E+2  7.1E+2  1.8E+1 -1.6E+2  6.9E-1 -8.4E+0 -7.5E+1 -2.1E+2 -6.6E+2',
'            -3.9E+1',
'0ITERATION NO.:    5    OBJECTIVE VALUE:   2506.10546953541        NO. OF FUNC. EVALS.:11',
' CUMULATIVE NO. OF FUNC. EVALS.:       65',
' PARAMETER:  6.2E-2  1.1E+0 -1.0E+0 -1.1E+0  4.2E-1  1.2E-1  2.7E-1  7.1E-1  3.0E-1  1.3E-1',
'            -1.2E-1',
' GRADIENT:   8.6E+0  5.1E+1  2.9E+1 -3.5E+1  1.2E+1  2.2E+0 -2.5E+0  1.9E+1 -1.1E+1 -4.1E+0',
'            -1.3E+2',
'0ITERATION NO.:   10    OBJECTIVE VALUE:   2494.92785884531        NO. OF FUNC. EVALS.:12',
' CUMULATIVE NO. OF FUNC. EVALS.:      123',
' PARAMETER: -1.2E-1  9.9E-1 -1.1E+0 -1.2E+0  1.9E-1  1.6E-1  1.7E+0  5.0E-1  3.3E-1  1.2E-1',
'            -9.8E-2',
' GRADIENT:   1.2E+1  3.1E+1 -3.1E+1 -1.5E+1  1.9E-1  5.3E+0  3.2E+0  4.0E-1 -3.6E+0 -3.3E+0',
'            -8.1E+1',
'0ITERATION NO.:   15    OBJECTIVE VALUE:   2487.99745856896        NO. OF FUNC. EVALS.:11',
' CUMULATIVE NO. OF FUNC. EVALS.:      178',
' PARAMETER: -9.0E-2  9.8E-1 -1.0E+0 -1.0E+0  1.8E-1  1.0E-1  1.3E+0  4.9E-1  2.8E-1  1.8E-1',
'            -4.7E-2',
' GRADIENT:  -5.0E-1  2.2E-1 -1.7E+0  7.8E-1  6.4E-1 -1.3E-1 -1.1E-1  1.7E-1  1.7E-1  1.4E+0',
'             2.8E-1',
'0ITERATION NO.:   20    OBJECTIVE VALUE:   2487.98367110010        NO. OF FUNC. EVALS.:21',
' CUMULATIVE NO. OF FUNC. EVALS.:      246',
' PARAMETER: -8.8E-2  9.8E-1 -1.0E+0 -1.0E+0  1.7E-1  1.0E-1  1.3E+0  4.9E-1  2.8E-1  1.7E-1',
'            -4.7E-2',
' GRADIENT:  -5.8E-1 -1.1E+0 -4.6E+0 -4.7E-1 -2.2E-1 -1.2E-1 -2.9E-1  4.8E-2  1.5E-1 -5.1E-1',
'            -1.3E-1',
'0ITERATION NO.:   25    OBJECTIVE VALUE:   2487.88855559751        NO. OF FUNC. EVALS.:16',
' CUMULATIVE NO. OF FUNC. EVALS.:      340',
' PARAMETER: -8.8E-2  1.0E+0 -9.9E-1 -1.0E+0  1.7E-1  9.7E-2  1.4E+0  4.8E-1  2.7E-1  1.8E-1',
'            -4.6E-2',
' GRADIENT:  -9.4E-3  8.1E-3  4.5E-2 -1.9E-2 -7.8E-4 -1.0E-3 -7.2E-4  3.5E-3 -2.6E-3  1.1E-2',
'             9.0E-3',
' Elapsed estimation time in seconds:     3.11',
' ',
' #TERM:',
'0MINIMIZATION SUCCESSFUL',
' NO. OF FUNCTION EVALUATIONS USED:      340',
' NO. OF SIG. DIGITS IN FINAL EST.:  3.2',
'',
' ETABAR IS THE ARITHMETIC MEAN OF THE ETA-ESTIMATES,',
' AND THE P-VALUE IS GIVEN FOR THE NULL HYPOTHESIS THAT THE TRUE MEAN IS 0.',
'',
' ETABAR:         2.3E-3  8.5E-4  6.8E-4',
' SE:             6.8E-2  4.6E-2  4.7E-2',
'',
' P VAL.:         9.7E-1  9.8E-1  9.8E-1',
' ',
' ETAshrink(%):   7.3E-1  1.6E+1  6.5E+0',
' EPSshrink(%):   8.8E+0',
' ',
' #TERE:',
' Elapsed covariance time in seconds:     4.27',
'1'
)
iterations <- iterations(lst)
iterations
it.dat <- melt(iterations,measure.var=names(iterations)[contains('X',names(iterations))])
xyplot(value~iteration|variable,it.dat[it.dat$course=='parameter',],
type='l',ylab='scaled parameter',as.table=TRUE,scales=list(y=list(relation='free')))
xyplot(value~iteration|variable,it.dat[it.dat$course=='gradient',] ,
type='l',ylab='gradient',as.table=TRUE,scales=list(y=list(relation='free')))

metrumrg documentation built on May 2, 2019, 5:55 p.m.