## ----include = FALSE----------------------------------------------------------
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

## ----message = FALSE, warning = FALSE-----------------------------------------
library(insurancerating)

portfolio <- as.data.frame(MTPL2)
portfolio$area <- factor(portfolio$area)
portfolio$area <- set_reference_level(portfolio$area, portfolio$exposure)

intercept_model <- glm(
  nclaims ~ 1 + offset(log(exposure)),
  family = poisson(),
  data = portfolio
)

area_model <- glm(
  nclaims ~ area + offset(log(exposure)),
  family = poisson(),
  data = portfolio
)

## -----------------------------------------------------------------------------
model_performance(intercept_model, area_model)

## -----------------------------------------------------------------------------
area_table <- rating_table(
  area_model,
  model_data = portfolio,
  exposure = "exposure"
)

area_table

area_review <- area_table |>
  add_portfolio_experience(
    data = portfolio,
    claim_count = "nclaims",
    exposure = "exposure",
    metric = "frequency"
  )

autoplot(
  area_review,
  risk_factors = "area",
  metric = "frequency"
)

## -----------------------------------------------------------------------------
set.seed(123)

bootstrap_result <- bootstrap_performance(
  area_model,
  portfolio,
  n_resamples = 50,
  sample_fraction = 0.8,
  sampling = "bootstrap",
  show_progress = FALSE
)

autoplot(bootstrap_result)

## -----------------------------------------------------------------------------
dispersion_check <- check_overdispersion(area_model)
dispersion_check

## ----message = FALSE, warning = FALSE-----------------------------------------
set.seed(123)

residual_check <- check_residuals(
  area_model,
  n_simulations = 250
)

autoplot(residual_check)

## -----------------------------------------------------------------------------
validation_data <- portfolio
validation_data$expected_claims <- predict(
  area_model,
  newdata = validation_data,
  type = "response"
)

area_oe <- rating_grid(
  validation_data,
  group_by = "area",
  exposure = "exposure",
  aggregate_cols = c("nclaims", "expected_claims")
)

area_oe$observed_frequency <-
  area_oe$nclaims / area_oe$exposure
area_oe$expected_frequency <-
  area_oe$expected_claims / area_oe$exposure
area_oe$observed_expected_ratio <- ifelse(
  area_oe$expected_claims > 0,
  area_oe$nclaims / area_oe$expected_claims,
  NA_real_
)

area_oe