Skip to contents

fit_bam_multi Fit a Generalized Additive Model using bam() with factor-smooth interactions to butterfly count data across multiple years simultaneously. This function fits a single model across multiple years using factor-smooth interactions (fs), which allows year-specific smooth terms while sharing information across years. This is more efficient than fitting separate models per year.

Source: R/fit_bam_multi.r
fit_bam_multi.Rd

fit_bam_multi Fit a Generalized Additive Model using bam() with factor-smooth interactions to butterfly count data across multiple years simultaneously. This function fits a single model across multiple years using factor-smooth interactions (fs), which allows year-specific smooth terms while sharing information across years. This is more efficient than fitting separate models per year.

Usage

fit_bam_multi(
  dataset_multi,
  NbrSample = NULL,
  GamFamily = "poisson",
  MaxTrial = 4,
  TimeUnit = "d",
  MultiVisit = "mean",
  weekday = 3,
  tp_col = NULL,
  smooth_basis = "cr",
  k_value = -1,
  verbose = TRUE,
  ...
)

Arguments

dataset_multi

data.table Butterfly counts for species x over multiple years y over all sites.

NbrSample

integer Maximum number of sites to sample per year, default=NULL (use all sites).

GamFamily

string Family for GAM, default='poisson', but can be 'nb' or 'quasipoisson'.

MaxTrial

integer Maximum number of fitting attempts, default=4.

TimeUnit

character Time-step for the spline, 'd' for day or 'w' for week.

MultiVisit

string Function for summarising multiple counts within a time unit, 'max' or 'mean' (default).

weekday

Integer for selected day of the week for weekly summary, default is 3 (Wednesday). [1-7] where 1 = Monday.

tp_col

string Name of temporal variable used in the GAM model, default NULL.

smooth_basis

string Basis for the smooth term, default "cr" (cubic regression spline). Other options include "tp" (thin plate), "ps" (P-splines).

k_value

integer Basis dimension for the smooth, default=-1 (automatic selection).

verbose

logical Indicating if progress reports should be given.

...

Additional parameters passed to bam function from the bam package.

Value

A list with three objects: i) **f_curve**: a data.table with the flight curves for all years with expected relative abundance (NM), normalized to sum to one per year, ii) **f_model**: the resulting bam model fitted on the count data across all years, iii) **f_data**: a data.table with the data used to fit the BAM model.

Details

This function uses factor-smooth interactions (s(tp_col, M_YEAR, bs="fs")) which allows: - Fitting a single model for multiple years (more efficient) - Year-specific smooths that can differ in shape - Sharing of information across years for more stable estimates - Better handling of years with sparse data

See also

flight_curve_multi, fit_gam, bam

Author

Reto Schmucki - retoshm@ceh.ac.uk