ddf function constraints documentation needs clarifying

[lenthomas]

In ddf it is possible to add lower and upper bounds to the detection function parameters. However, I believe these just apply to the dsmodel parameters, not the mrmodel. The documentation for the ddf function does not make this clear:

lowerbounds
a vector of lowerbounds for the parameters

upperbounds
a vector of upperbounds for the parameters

I also think the initial values are only for the dsmodel.

These things need checked and the help clarified. This would be help for the ddf function and also the text in the mrds-opt page.

[LHMarshall]

Code checking

No control values are passed to the model fitting function (io.glm) from ddf.io.fi

Check functionality by running

data(book.tee.data)
region <- book.tee.data$book.tee.region
egdata <- book.tee.data$book.tee.dataframe
samples <- book.tee.data$book.tee.samples
obs <- book.tee.data$book.tee.obs

# Confirm how the initial values work for the ds models

# fit a half-normal detection function
result <- ddf(dsmodel=~mcds(key="hn", formula=~1), data=egdata, method="ds",
              meta.data=list(width=4))
result$par

# fit a half-normal detection function
result <- ddf(dsmodel=~mcds(key="hn", formula=~1), data=egdata, method="ds",
              meta.data=list(width=4),
              control = list(initial = list(scale = 0.25)))
result$par # slightly different

result <- ddf(dsmodel=~mcds(key="hn", formula=~1), data=egdata, method="ds",
              meta.data=list(width=4),
              control = list(initial = list(scale = 0.25),
                             lowerbounds = 0, 
                             upperbounds = 0.9))
result$par # hits upper bound

# Seem to have no effect on the mr models

result <- ddf(dsmodel=~mcds(key="hn", formula=~1), data=egdata, method="ds",
              meta.data=list(width=4),
              control = list(initial = list(scale = 1.25),
                             lowerbounds = 1, 
                             upperbounds = 1.9))
result$par # hits lower bound

# fit an independent observer model with full independence
result.io.fi <- ddf(mrmodel=~glm(~distance), data=egdata, method="io.fi",
                    meta.data=list(width = 4))
result.io.fi$par

# fit an independent observer model with full independence
result.io.fi2 <- ddf(mrmodel=~glm(~distance), data=egdata, method="io.fi",
                    meta.data=list(width = 4),
                    control = list(lowerbounds = c(2,-1.1), upperbounds = c(2.5,-1)))

# No difference
result.io.fi$par
result.io.fi2$par

# fit an independent observer model with point independence
result.io <- ddf(dsmodel=~cds(key = "hn"), mrmodel=~glm(~distance),
                 data=egdata, method="io", meta.data=list(width=4))
result.io$par

result.io <- ddf(dsmodel=~cds(key = "hn"), mrmodel=~glm(~distance),
                 data=egdata, method="io", meta.data=list(width=4),
                 control = list(initial = list(scale = 0.25),
                                lowerbounds = 0, 
                                upperbounds = 0.9))
result.io$par #scale param of DF is 0.9

[LHMarshall]

Checking the order of the lower and upper bounds for covariates and adjustment terms

result <- ddf(dsmodel=~mcds(key="hr", formula=~size+sex+exposure), 
              data=egdata, method="ds",
              meta.data=list(width=4),
              control = list(initial = list(scale = c(0.5, 0.05, 0.5, 0.1),
                                            shape = 1.35)))
summary(result)
result$par

result2 <- ddf(dsmodel=~mcds(key="hr", formula=~size+sex+exposure), 
              data=egdata, method="ds",
              meta.data=list(width=4),
              control = list(initial = list(scale = c(0.5, 0.05, 0.5, 0.1),
                                            shape = 1.35),
                             lowerbounds = c(0.5, 0.4, 0,   0.4, 0  ),
                             upperbounds = c(1,   0.55, 0.1, 0.59, 0.069)))

result2$par
result$par

# Check where adjustment terms fit in

result3 <- ddf(dsmodel=~mcds(key="hr", formula=~size+sex+exposure, 
                             adj.series="cos", adj.order=2), 
               data=egdata, method="ds",
               meta.data=list(width=4))

> result$par
          V1 X.Intercept.         size          sex 
  1.35393807   0.51176985   0.05595641   0.50309413 
    exposure 
  0.11589361 
> result3$par
          V1 X.Intercept.         size          sex 
  1.51183107   0.64483879   0.03699285   0.40526429 
    exposure           V6 
  0.09326986   0.08810099 

[LHMarshall]

@lenthomas what is the mrds-opt page you refer to above?

[erex]

@LHMarshall There is a man-page written containing optimisation hints as seen here looking at the content of the mrds package, yet it cannot be accessed (as you have shown) by using the question mark. Don't know why the question mark doesn't work.

[LHMarshall]

@erex it had a problem with the - in mrds-opt so I've renamed to mrds_opt now ?mrds_opt works