Skip to contents

Assess timeseries for trends and status

Source: R/assessment_functions.R
run_assessment.Rd

Fits a model to each timeseries, test for any temporal trend and compare with thresholds. Need to add a lot more in details.

Usage

run_assessment(
  ctsm_ob,
  subset = NULL,
  AC = NULL,
  get_AC_fn = NULL,
  recent_trend = 20L,
  parallel = FALSE,
  extra_data = NULL,
  control = list(),
  ...
)

Arguments

ctsm_ob

A HARSAT object resulting from a call to create_timeSeries

subset

An optional vector specifying which timeseries are to be assessed. Might be used if the assessment is to be done in chunks because of size, or when refitting a timeseries model which has not converged. An expression will be evaluated in the timeSeries component of ctsm_ob; use 'series' to identify individual timeseries.

AC

A character vector identifying the thresholds to be used in status assessments. These should be in the threshold reference table. Defaults to NULL; i.e. no thresholds are used.

get_AC_fn

An optional function that overrides get_AC_default. See details (which need to be written).

recent_trend

An integer giving the number of years which are used in the assessment of recent trends. For example, a value of 20 (the default) consider trends in the last twenty year.

parallel

A logical which determines whether to use parallel computation; default = FALSE.

extra_data

[Experimental] A named list used to pass additional data to specific assessment routines. At present it is only used for imposex assessments, where it passes two data frames called VDS_estimates and VDS_confidence_limits. Defaults to NULL, This argument will be generalised in the near future, so expect it to change.

control

[Experimental] A list of control parameters that allow the user to modify the way the assessment is run. These currently include parameters for post-hoc power calculations and for the calculation of the 'recent_change' (in addition to the recent_trend argument above). See details.

...

Extra arguments which are passed to assessment_engine. See details (which need to be written).

Details

Control parameters

Some aspects of the model output are controlled using parameters which can be modified using the control argument. The default parameter list is described below. The control argument only needs to specify any changes. For example, to change the target power from 90% (default) to 80%, use

control = list(power = list(target_power = 80))

The default parameters are a list with the following components:

power

A list with the following components (all expressed as percentages):

  • target_power default = 90

  • target_trend default = 5

  • size default = 5

These affect the post-how power calculations. Power is currently only calculated for time series of log-normally distributed data, which is why the trend is expressed as a percentage.

recent_change

A list with the following components:

  • n_year_fit default = 5L

  • n_year_positive default = 5L

A recent change will only be computed if there are at least n_year_fit years of data in the recent period, of which at least n_year_positive contain at least one non-censored measurement. This only affects normally or log-normally distributed data.

The default values of 5L mirror the requirements for calculating the change over the whole time series.

Note that in harsat versions <= 1.0.2, n_year_positive was hard-wired to 0L which occasionally led to pathological behaviour in the estimation of the recent change. To replicate previous outputs as closely as possible, set n_year_positive to 2L (the smallest value allowed to avoid any pathologial behavour). The change is only likely to affect long time series with infrequent sampling where most measurements in the recent period are censored.