summary.simulations: Generic Methods for secrdesign Objects

Description Usage Arguments Details Value See Also Examples

Description

Methods to summarize simulated datasets.

Usage

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
## S3 method for class 'secrdesign'
summary(object, ...)

## S3 method for class 'rawdata'
summary(object, ...)

## S3 method for class 'estimatetables'
summary(object, ...)

## S3 method for class 'selectedstatistics'
summary(object, fields = c('n', 'mean',
'se'), dec = 5, alpha = 0.05, type = c('list','dataframe','array'), ...)

## S3 method for class 'selectedstatistics'
plot(x, scenarios, statistic, type =
c('hist', 'CI'), refline, xlab = NULL, ...)

header(object)

Arguments

object

object of class simulations from run.scenarios

dec

number of decimal places in output

fields

character vector; names of required summary statistics (see Details)

alpha

alpha level for confidence intervals and quantiles

type

character code for type of output (see Details)

...

other arguments – not currently used by summary but passed to hist by the plot method

x

object of class ‘selectedstatistics’ from run.scenarios

scenarios

integer indices of scenarios to plot (all plotted if not specified)

statistic

integer or character indices if the statistics in x for which histograms are requested

refline

logical; if TRUE a reference line is plotted at the true value of a parameter

xlab

character; optional label for x-axis

Details

If object inherits from ‘selectedstatistics’ then the numeric results from replicate simulations are summarized using the chosen ‘fields’ (by default, the number of non-missing values, mean and standard error), along with header information describing the simulations. Otherwise the header alone is returned.

fields is a vector of any selection from c(‘n’, ‘mean’, ‘sd’, ‘se’, ‘min’, ‘max’, ‘lcl’, ‘ucl’, ‘median’, ‘q’, ‘rms’), or the character value ‘all’.

Field ‘q’ provides 1000 alpha/2 and 1000[1 - alpha/2] quantiles qxxx and qyyy.

‘lcl’ and ‘ucl’ refer to the upper and lower limits of a 100(1 - alpha)% confidence interval for the statistic, across replicates.

‘rms’ gives the root-mean-square of the statistic - most useful for the statistic ‘ERR’ (see select.stats) when it represents the overall accuracy or RMSE.

The plot method plots either (i) histograms of the selected statistics (type = ‘hist’) or (ii) the estimate and confidence interval for each replicate (type = ‘CI’). The default for type = ‘hist’ is to plot the first statistic - this is usually ‘n’ (number of detected animals) when fit = FALSE, and ‘estimate’ (parameter estimate) when fit = TRUE. If length(statistic) > 1 then more than one plot will be produced, so a multi-column or multi-row layout should be prepared with par arguments ‘mfcol’ or ‘mfrow’.

For type = ‘CI’ the statistics must include ‘estimate’, ‘lcl’ and ‘ucl’ (or ‘beta’, ‘lcl’ and ‘ucl’ if outputtype = ‘coef’).

Value

List with components ‘header’

call

original function call

starttime

from object

proctime

from object

constants

small dataframe with values of non-varying inputs

varying

small dataframe with values of varying inputs

fit.args

small dataframe with values arguments for secr.fit, if specified

and ‘OUTPUT’, a list with one component for each field. Each component may be a list or an array.

See Also

run.scenarios, make.array, select.stats validate

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
## collect raw counts
scen1 <- make.scenarios(D = c(5,10), sigma = 25, g0 = 0.2)
traps1 <- make.grid()
tmp1 <- run.scenarios(nrepl = 50, trapset = traps1, scenarios = scen1,
    fit = FALSE)

opar <- par(mfrow=c(2,3))
plot(tmp1, statistic = 1:3)
par(opar)

summary(tmp1)

summary(tmp1, field=c('q025', 'median', 'q975'))

secrdesign documentation built on Oct. 10, 2017, 5:06 p.m.