Data_Fn: Data inputs for TMB

In aaronmberger/Geo_dGLMM_habitat: Package for conducting spatial estimation of CPUE index standardization (fisheries, ecology, etc).

Description

Generates a tagged list representing data inputs for use by TMB, and tries to check validity of inputs

Usage

Data_Fn(Aniso, FieldConfig, ObsModel, Options, b_i, a_i, v_i, s_i, t_i, a_xl, X_xj, Q_ik, MeshList)

Arguments

Aniso	Use geometric anisotropy for spatial variation (0=no, 1=yes)
FieldConfig	Vector describing what types of spatial variation to estimate
ObsModel	Integer for distribution of residual 'measurement' error (0=normal (log-link); 1=lognormal; 2=gamma; 4=ZANB; 5=ZINB; 11=lognormal-mixture; 12=gamma-mixture)
Options	Essentially deprecated
b_i	biomass for each observation i
a_i	Area swept (or other offset) for each observation i
v_i	vessel for each observation
s_i	Polygon associated with each observation
t_i	Year for each observation
a_xl	Matrix with dimension n_x by n_l, where a_xl[x,l] gives the total area in polygon x that should be included when calculating index l. Indices do not need to be mutually exclusive (i.e., rows of a_xl can sum to more than one)
X_xj	Matrix with dimension n_x by n_j, where X_xj[x,j] gives the value of the j-th covariate for polygon x
Q_ik	Matrix with dimension n_i by n_k, where Q_ik[i,k] gives the value of 'catchability' covariate k for sample i
MeshList	named list generated by Calc_Anisotropic_Mesh() function

Details

This function generates the named list that is then used by MakeADFun() to generate the TMB object.

Value

There are three matrix inputs (a_xl, X_xj, and Q_ik) that require specification:

a_xl	a matrix used when post-processing results to calculate indices that are informed by subsets of the data. Using a conventional area-weighting approach, the l-th index is calculated as the sum of a_xl[x,l]*D_x[x], where D_x[x] is the density in each polygon x. For example, if a_xl is a vector of 1s, then all polygons are calculated as having equal weight when calculating an index.
X_xj	a matrix of j measured covariates that contribute to spatial variation in density. These covariates are then used when calculating density D_x for each polygon x.
Q_ik	a matrix of k covariates that represent variation in catch rates but not variation in density, e.g., gear efficient and catchability. This could include vessel horsepower, bottom temperature, etc. These variables are NOT used when calculating D_x for each polygon x. Instead, D_x is calculated for a reference value of Q_ik = 0 (i.e., when controling for variables in Q_ik).

This function also calculates five dimensions that are needed in TMB:

n_t	Number of time periods (typically years) in data set
n_v	Number of vessels in data set
n_i	the number of observations in the data set
n_x	the number of polygons used to summarize spatial variation
n_s	the number of vertices ('knots') in the triangulated mesh calculated by Calc_Anisotropic_Mesh()
n_j	the number of covariates for spatial variation
n_k	the number of covariates describing variation in catchability among samples
n_l	the number of indices to be calculated via post-processing results

aaronmberger/Geo_dGLMM_habitat documentation built on May 10, 2019, 3:20 a.m.