Run glm modeling and forecasting — glm

This is a wrapper function that pipelines influenza hospitalization modeling (glm_fit) and forecasting (glm_forecast).

Usage

glm_wrap(
  .data,
  .models,
  new_covariates = NULL,
  horizon = 4,
  alpha = c(0.01, 0.025, seq(0.05, 0.45, by = 0.05)) * 2
)

Arguments

.data: Data including all explanatory and outcome variables needed for modeling
.models: List of models defined as trending::trending_model objects
new_covariates: A tibble with one column per covariate, and n rows for n horizons being forecasted
horizon: Number of weeks ahead for forecasting
alpha: Vector specifying the threshold(s) to be used for prediction intervals (PI); alpha of 0.05 would correspond to 95% PI; default is c(0.01, 0.025, seq(0.05, 0.45, by = 0.05)) * 2 to create a range of intervals

Value

Named list with two elements:

model: Output from glm_fit with selected model fit
forecasts: Output from glm_forecast with forecasts from each horizon combined as a single tibble

Examples

if (FALSE) { # \dontrun{
# Retrieve data to be used in fitting models
hosp_va <-
 get_hdgov_hosp(limitcols=TRUE) %>%
 prep_hdgov_hosp(statesonly=TRUE, min_per_week = 0, remove_incomplete = TRUE) %>%
 dplyr::filter(abbreviation == "VA")

# Define list of models
models <-
 list(
   poisson = trending::glm_model(flu.admits ~ hosp_rank + ili_rank, family = "poisson"),
   quasipoisson = trending::glm_model(flu.admits ~ hosp_rank + ili_rank, family = "quasipoisson"),
   negbin = trending::glm_nb_model(flu.admits ~ hosp_rank + ili_rank)
 )

# Create new covariate data to feed into forecast procedure
new_cov <-
  dplyr::tibble(
    date = max(hosp_va$week_start) + c(7,14,21,28),
    epiweek = lubridate::epiweek(date),
    epiyear = lubridate::epiyear(date)
  ) %>%
  dplyr::left_join(
    fiphde:::historical_severity, by="epiweek"
  ) %>%
  dplyr::select(-epiweek,-epiyear)

# Run the glm wrapper to fit and forecast
va_glm_res <- glm_wrap(.data = hosp_va, .models = models, new_covariates = new_cov, horizon = 4)
va_glm_res

} # }