Join predictive distributions into a portfolio

join_predictive() puts distributions of different models side by side (a reserve bootstrap by origin, premium risk by line, a tower's net), with a new leading key column dim holding each part's name, followed by the union of the parts' key columns ("" where a part lacks one). Simulation i of the result is simulation i of every part. reorder_groups() then sets the dependence between the groups of dim by Iman-Conover on the group totals, moving each group's simulations as whole rows: every group keeps its distribution and internal joint structure. Use aggregate(), VaR(), TVaR() and capital_allocation() on the result.

Usage

join_predictive(parts, dim, same_simulations = FALSE)

reorder_groups(x, dim, correlation, seed)

Arguments

parts: A named list of predictive_distributions with the same number of simulations.
dim: Name of the new key column.
same_simulations: FALSE: the parts were simulated separately, and two with the same seed (which would share random numbers) are refused. TRUE: the parts come from the same scenarios (a cover applied to a reserve) and keep their pairing.
x: A predictive_distribution.
correlation: Correlation matrix, one row and column per group in order of first appearance.
seed: Seed, a whole number.

Value

A predictive_distribution.

Examples

a <- predictive_distribution(cbind(c(10, 12), c(20, 25)), data.frame(origin = c(2023, 2024)))
b <- predictive_distribution(matrix(c(50, 40), ncol = 1), data.frame(lob = "motor"))
p <- join_predictive(list(reserve = a, premium = b), "risk")
p@keys
#>      risk origin   lob
#> 1 reserve   2023      
#> 2 reserve   2024      
#> 3 premium        motor