Calibrated synthetic long-format loss experience data for
demonstrating the lossratio package workflow. Generated
deterministically from data-raw/make_experience.R (set.seed =
20260501). The marginal LR-by-development curve for each of the four
coverages is calibrated to the broad shape of a real long-term
insurance portfolio's first ten months of development; the elap_m
11-30 plateau, all cohort-level patterns, demographic mixes, and
cell-level loss / rp values are randomly drawn. The SUR coverage
carries a synthetic 2024-04 cohort regime break (LR halved) so the
detect_cohort_regime() example has a clear shift to find.
Format
A data.table with 33,480 rows and 17 columns:
- cy, cyh, cyq, cym
Calendar period (year / half / quarter / month) as
Date.- uy, uyh, uyq, uym
Underwriting period (year / half / quarter / month) as
Date.- elap_y, elap_h, elap_q, elap_m
Elapsed period (year / half / quarter / month) as integer.
- cv_nm
Coverage name (character).
- age_band
Age band (ordered factor).
- gender
Gender (factor).
- loss
Loss amount (perturbed).
- rp
Risk premium amount (perturbed).
Source
Generated by data-raw/make_experience.R. Per-coverage
marginal LR-by-development calibrated to broad real-portfolio
aggregates; cell-level structure and other patterns are random.
Examples
if (FALSE) { # \dontrun{
data(experience)
head(experience)
exp <- as_experience(experience)
tri <- build_triangle(exp, group_var = cv_nm)
} # }