density_bins: Density bins as data frames suitable for use with predict_curve

Description

Generates a data frame of bins representing the kernel density (or histogram) of a vector, suitable for use in generating predictive distributions using predict_curve.

Usage

density_bins(x, n = 101, ...)
histogram_bins(x, n = 30, breaks = n, ...)

Arguments

A numeric vector

Number of bins

...

Additional arguments passed to density or hist.

breaks

Used to set bins for histogram_bins. Can be number of bins (by default it is set to the value of n) or a method for setting bins. See the breaks argument of hist.

Value

A data frame representing bins and their densities with the following columns:

mid

Bin midpoint

lower

Lower endpoint of each bin

upper

Upper endpoint of each bin

density

Density estimate of the bin

Details

These functions are simple wrappers to density and hist that compute density estimates and return their results in a consistent format: a data frame of bins suitable for use with predict_curve.

density_bins computes a kernel density estimate using density.

histogram_bins computes a density histogram using hist.

Examples

Run this code

# NOT RUN {
library(ggplot2)
library(dplyr)
library(purrr)
library(tidyr)

if (
  require("rstanarm", quietly = TRUE) &&
  require("modelr", quietly = TRUE)
) {

  theme_set(theme_light())

  m_mpg = stan_glm(mpg ~ hp * cyl, data = mtcars)

  step = 1
  mtcars %>%
    group_by(cyl) %>%
    data_grid(hp = seq_range(hp, by = step)) %>%
    add_predicted_draws(m_mpg) %>%
    summarise_all(list) %>%
    mutate(densities = map(.prediction, density_bins)) %>%
    unnest(densities) %>%
    ggplot() +
    geom_rect(aes(
      xmin = hp - step/2, ymin = lower, ymax = upper, xmax = hp + step/2,
      fill = ordered(cyl), alpha = density
    )) +
    geom_point(aes(x = hp, y = mpg, fill = ordered(cyl)), shape = 21, data = mtcars) +
    scale_alpha_continuous(range = c(0, 1)) +
    scale_fill_brewer(palette = "Set2")
}
# }