Fit chain ladder projection from a Triangle object

Fit a chain ladder projection from an object of class "Triangle". The function works on long-form cumulative data and does not require a complete triangle.

Two methods are supported via the method argument:

"basic" (default): Classical chain ladder point projection. Age-to-age factors are estimated through build_ata() and fit_ata(), then applied recursively.
"mack": Mack (1993) chain ladder. Point forecast follows the standard recursion, and prediction uncertainty is decomposed into process variance and parameter variance.

When weight_var is supplied (e.g. "crp"), age-to-age factors and their variance are estimated using the supplied WLS weights.

Usage

fit_cl(
  x,
  method = c("basic", "mack"),
  value_var = "closs",
  weight_var = NULL,
  alpha = 1,
  sigma_method = c("min_last2", "locf", "loglinear"),
  recent = NULL,
  maturity_args = NULL,
  tail = FALSE
)

Arguments

x: An object of class "Triangle".
method: One of "basic" or "mack". Default is "basic".
value_var: A single cumulative variable to project. Typical choices are "closs", "crp", or "clr".
weight_var: An optional column name passed to build_ata() as the WLS weight variable. Typically "crp" when value_var = "clr". Default is NULL.
alpha: Numeric scalar controlling the variance structure in fit_ata(). Default is 1.
sigma_method: Sigma extrapolation method passed to fit_ata(). One of "min_last2" (default), "locf", or "loglinear". Only relevant when method = "mack".
recent: Optional positive integer. When supplied, only the most recent recent periods are used for factor estimation. Default is NULL (use all periods).
maturity_args: A named list of arguments forwarded to find_ata_maturity() via fit_ata(), or NULL (default) to skip maturity filtering. Pass list() to use all defaults with maturity filtering enabled.
tail: Logical or numeric. If FALSE, no tail factor is applied. If TRUE, a log-linear tail factor is estimated from selected factors. If numeric, the supplied value is used as the tail factor.

Value

An object of class "CLFit" containing:

call: The matched call.
data: The input "Triangle" object.
method: The method used ("basic" or "mack").
group_var: Character vector of grouping variable names.
cohort_var: Character scalar of period variable name.
dev_var: Character scalar of development variable name.
value_var: Character scalar of value variable name.
full: data.table with observed and projected values. For "mack", also includes process/parameter SE and CV columns.
pred: data.table identical to full with observed cells set to NA.
ata: The "ATA" object produced by build_ata().
summary: For "basic": data.table of fitted factors from fit_ata(). For "mack": cohort-level summary with latest, ultimate, reserve, and Mack standard errors.
selected: data.table of selected factors used for projection.
factor: For "mack" only: data.table of fitted factors from fit_ata().
maturity: Maturity diagnostics from find_ata_maturity(), or NULL when maturity filtering was not applied.
alpha: Value of alpha used.
sigma_method: For "mack" only: sigma extrapolation method.
weight_var: Weight variable name used, or NULL.
recent: Number of recent periods used, or NULL.
use_maturity: Logical; whether maturity filtering was applied.
maturity_args: Resolved maturity arguments, or NULL.
tail: Tail factor argument supplied by the user.
tail_factor: Numeric tail factor applied.

Examples

if (FALSE) { # \dontrun{
data(experience)
exp <- as_experience(experience)
tri <- build_triangle(exp[cv_nm == "SUR"], group_var = cv_nm)

# Basic chain ladder (point projection only)
cl <- fit_cl(tri, value_var = "closs", method = "basic")
print(cl)

# Mack chain ladder with process / parameter standard errors
cl_mack <- fit_cl(tri, value_var = "closs", method = "mack")
summary(cl_mack)
plot(cl_mack)

# WLS factors for clr (loss ratio) using crp as the weight
cl_clr <- fit_cl(tri, value_var = "clr", weight_var = "crp")
} # }

Fit chain ladder projection from a `Triangle` object

Usage

Arguments

Value

See also

Examples