Estimation options

From DEBwiki
Jump to: navigation, search
AmP estimation
Concepts

Data and completeness
Parameter estimation
Goodness-of-fit: SMSE / MRE
AmP Literature

Practice - essentials

Starting an estimation for a new species
Setting initial parameter values
Setting weight coefficients
Computing implied properties
Submitting to the collection

Practice - extra modules
Code specification
User-defined files: run, mydata, pars_init, predict
Data: Zero-variate, Univariate, Pseudo-data

Typified models
Estimation options


The AmP estimation procedure allows for the choice of several different options using the function estim_options. All options and descriptions are summarized in the table below. For more detail on the loss function see next section. A table summarizing all default options is provided at the end of the page.


Options Description
pars_init_method

Parameter initialization method

  • 0 - get initial estimates from automatized computation(does not always work, more research is needed)
  • 1 - read initial estimates from .mat file (for continuation)
  • 2 - read initial estimates from pars_init file(default)
filter
  • 0 - do not use filter [the search will be unrestricted]
  • 1 - use filter (default) [the search will be performed in the volume of the parameter space defined by the boundaries]
loss_function
  • sb - symmetric bounded (default)
  • su - symmetric unbounded
  • re - relative error (original one, no longer in use since the DEB2017 release of DEBtool)
report
  • 1 - prints to screen the step numbers and the corresponding simplex ssq values (default)
  • 0 - does not print to screen the step numbers and the corresponding simplex ssq values
results_output

Results output format

  • 0 - only saves data to .mat (no printing to html or screen and no figures) - use this for (automatic) continuations
  • 1, -1 - no saving to .mat file, prints results to html (1) or screen (-1)
  • 2, -2 - saves to .mat file, prints results to html (2) or screen (-2)
  • 3, -3 - like 2 (or -2), but also prints graphs to .png files (default is 3)
  • 4, -4 - like 3 (or -3), but also prints html with implied traits
  • 5, -5 - like 3 (or -3), but also prints html with implied traits including related species
  • 6 - like 5, but also prints html with population traits

If a customized results file exists, the results output prints results using that file.

pseudodata

Pseudo-data choice for multispecies estimation

  • 0 - put pseudodata together with data for each pet (default)
  • 1 - put pseudodata apart for all pets (only for multispecies estimation)
method
  • 'nm' - use Nelder-Mead method (default)
  • 'no' - do not estimate
max_step_number maximum number of steps
max_fun_evals maximum number of function evaluations



Loss functions

Option Description Formula
sb symmetric bounded (by product) Lossfunction F.png
su symmetric unbounded (by product) Lossfunction J.png
re relative error (symmetric by addition) Lossfunction E.png

where i refers to the data set and j to a given point in data set i. dij stands for the data, pij for the model prediction and wij for the associated weight coefficient. Finally, di and pi in the denominator represent, respectively, the average of all data points (dij) and predictions (pij) in set i:

Mean d.png and Mean p.png


Summary of default options

Key Option Description
filter 1 the search will be performed in the volume of the parameter space defined by the boundaries
pars_init 2 read initial estimates from pars_init file
loss_function sb symmetric bounded (by product)
method nm use Nelder-Mead method
results 0 saves data to .mat file
pseudodata 0 put pseudodata together with data for each pet