Modeler

General-purpose optimization techniques for non-linear least squares problems.

Usage

modeler(
  data,
  x,
  y,
  grp,
  keep,
  fn = "fn_linear_sat",
  parameters = NULL,
  lower = -Inf,
  upper = Inf,
  fixed_params = NULL,
  method = c("subplex", "pracmanm", "anms"),
  subset = NULL,
  options = modeler.options(),
  control = list()
)

Arguments

data

A data.frame in a wide format.

x

The name of the column in `data` that contains x points.

y

The name of the column in `data` that contain the variable to be analyzed. Must match a var in the data.

grp

The names of the columns in `data` that contains a grouping variable. (Optional).

keep

The names of the columns in `data` to keep across the analysis.

fn

A string specifying the name of the function to be used for the curve fitting. Default is "fn_linear_sat".

parameters

Can be a named numeric vector specifying the initial values for the parameters to be optimized, or a data frame with columns uid, and the initial parameter values for each group id. Used for providing specific initial values per group id. Default is NULL.

lower

Numeric vector specifying the lower bounds for the parameters. Default is -Inf for all parameters.

upper

Numeric vector specifying the upper bounds for the parameters. Default is Inf for all parameters.

fixed_params

A data frame with columns uid, and the fixed parameter values for each group id. Used for fixing certain parameters during optimization.

method

A character vector specifying the optimization methods to be used. Check `optimx::checkallsolvers()` for available methods. Default is c("subplex", "pracmanm", "anms").

subset

An optional vector with levels of `grp` to filter the data. Default is NULL, meaning all groups are used.

options

A list of additional options. See `modeler.options()`

add_zero

Logical. If TRUE, adds a zero value to the series at the start. Default is FALSE.

check_negative

Logical. If TRUE, converts negative values in the data to zero. Default is FALSE.

max_as_last

Logical. If TRUE, appends the maximum value after reaching the maximum. Default is FALSE.

progress

Logical. If TRUE a progress bar is displayed. Default is FALSE. Try this before running the function: progressr::handlers("progress", "beepr").

parallel

Logical. If TRUE the model fit is performed in parallel. Default is FALSE.

workers

The number of parallel processes to use. `parallel::detectCores()`

trace

If TRUE , convergence monitoring of the current fit is reported in the console. FALSE by default.

return_method

Logical. If TRUE, includes the optimization method used in the result. Default is FALSE.

control

A list of control parameters to be passed to the optimization function. For example: list(maxit = 500).

Value

An object of class modeler, which is a list containing the following elements:

param

A data frame containing the optimized parameters and related information.

dt

A data frame with data used and fitted values.

fn

The call used when fitting models.

metrics

Metrics and summary of the models.

execution

Execution time.

response

Response variable.

keep

Metadata to keep across.

fun

Name of the function.

parallel

List returning parallel and workers.

fit

List with the fitted models.

Examples

library(flexFitR)
data(dt_potato)
explorer <- explorer(dt_potato, x = DAP, y = c(Canopy, GLI), id = Plot)
# Example 1
mod_1 <- dt_potato |>
  modeler(
    x = DAP,
    y = GLI,
    grp = Plot,
    fn = "fn_lin_pl_lin",
    parameters = c(t1 = 38.7, t2 = 62, t3 = 90, k = 0.32, beta = -0.01),
    subset = 195
  )
plot(mod_1, id = 195)

print(mod_1)
#> 
#> Call:
#> GLI ~ fn_lin_pl_lin(DAP, t1, t2, t3, k, beta) 
#> 
#> Residuals:
#>      Min.   1st Qu.    Median      Mean   3rd Qu.      Max. 
#> -0.011076 -0.001003  0.000000 -0.001886  0.000000  0.000000 
#> 
#> Optimization Results `head()`:
#>  uid   t1   t2   t3     k     beta      sse
#>  195 40.1 63.1 91.4 0.325 -0.00809 0.000139
#> 
#> Metrics:
#>  Groups      Timing Convergence Iterations
#>       1 0.7244 secs        100%  3156 (id)
#> 
# Example 2
mod_2 <- dt_potato |>
  modeler(
    x = DAP,
    y = Canopy,
    grp = Plot,
    fn = "fn_linear_sat",
    parameters = c(t1 = 45, t2 = 80, k = 0.9),
    subset = 195
  )
plot(mod_2, id = 195)

print(mod_2)
#> 
#> Call:
#> Canopy ~ fn_linear_sat(DAP, t1, t2, k) 
#> 
#> Residuals:
#>       Min.    1st Qu.     Median       Mean    3rd Qu.       Max. 
#> -1.201e-09  0.000e+00  1.911e-10  1.031e-10  3.823e-10  8.791e-10 
#> 
#> Optimization Results `head()`:
#>  uid   t1   t2   k      sse
#>  195 38.8 61.6 100 2.65e-18
#> 
#> Metrics:
#>  Groups      Timing Convergence Iterations
#>       1 0.4516 secs        100%   351 (id)
#>