inst/doc/GetStarted.R

In bayesian: Bindings for Bayesian TidyModels

## ----results='hide', message=FALSE, warning=FALSE-----------------------------
library(bayesian)

## ----results='hide', message=FALSE, warning=FALSE-----------------------------
library(recipes)
library(workflows)

## -----------------------------------------------------------------------------
epi_recipe <- epilepsy |>
  recipe() |>
  update_role(count, new_role = "outcome") |>
  update_role(Trt, Age, Base, patient, new_role = "predictor") |>
  add_role(patient, new_role = "group") |>
  step_normalize(Age, Base)

## -----------------------------------------------------------------------------
print(epi_recipe)

## -----------------------------------------------------------------------------
epi_model <- bayesian(
    family = poisson()
  ) |>
  set_engine("brms") |>
  set_mode("regression")

## -----------------------------------------------------------------------------
print(epi_model)

## -----------------------------------------------------------------------------
epi_model <- epi_model |>
  update(family = poisson())

## -----------------------------------------------------------------------------
epi_workflow <- workflow() |>
  add_recipe(epi_recipe) |>
  add_model(
    spec = epi_model,
    formula = count ~ Trt + Base + Age + (1 | patient)
  )

## -----------------------------------------------------------------------------
print(epi_workflow)

## ----results='hide', echo = FALSE---------------------------------------------
run_on_linux <- grepl("linux", R.Version()$os, ignore.case = TRUE)

## ----results='hide', eval = run_on_linux--------------------------------------
epi_workflow_fit <- epi_workflow |>
  fit(data = epilepsy)

## ----eval = run_on_linux------------------------------------------------------
print(epi_workflow_fit)

## ----eval = run_on_linux------------------------------------------------------
epi_fit <- epi_workflow_fit |>
  extract_fit_parsnip()

## ----eval = run_on_linux------------------------------------------------------
epi_brmsfit <- epi_workflow_fit |>
  extract_fit_engine()

## ----eval = run_on_linux------------------------------------------------------
class(epi_brmsfit)

## -----------------------------------------------------------------------------
newdata <- epilepsy[1:5, ]

## ----eval = run_on_linux------------------------------------------------------
epi_workflow_fit |>
  predict(
    new_data = newdata,
    type = "conf_int",
    level = 0.95
  )

## ----eval = run_on_linux------------------------------------------------------
epi_workflow_fit |>
  predict(
    new_data = newdata,
    type = "conf_int",
    level = 0.95,
    std_error = TRUE
  )