R/helpers.R

In ericnovik/introstats: Data and Functions for Statisitcs With R Book by Stinerock and Novik

#' Updates the introstats pacakge from github
#'
#' @examples
#' update_package()
update_package <- function() {
  if (suppressMessages(suppressWarnings(require("introstats")))) {
    remove.packages("introstats")
  }
  if (!suppressMessages(suppressWarnings(require("devtools")))) {
    install.packages("devtools")
  }
  devtools::install_github("ericnovik/introstats")
}

#' Plots normal densities
#'
#' @param n_dens Number of densities to draw with standard deviations from 1 to n_dens
#' @return ggplot2 plot object
#' @examples
#' plot_normals()
#' plot_normals(5)
plot_normals <- function (n_dens = 3) {
  n <- 1000
  x <- seq(-10, 10, length.out = n)
  dens <- matrix(ncol = n_dens, nrow = n)

  for (i in 1:n_dens) {
    dens[, i] <- dnorm(x, mean = 0, sd = i)
  }

  colnames(dens) <- paste0("sd_", 1:n_dens)
  dens <- as.data.frame(dens)
  dens <- dplyr::mutate(dens, z_score = x)
  dens <- tidyr::gather(dens, sd, obs, -z_score)
  ggplot2::qplot(z_score, obs, data = dens, colour = sd, geom = "line") +
    ggplot2::ylab("Density") +
    ggplot2::xlab("Z Score") +
    ggplot2::theme_bw()
}

#' Plots the distributions of the sample means from the population.
#'
#' @param pop Vector containing the population (anything over 10,000 will take a long time to run on most ocmputers)
#' @param n Number of replicas of each sample
#' @param size Vector containing sizes for each sampling distribution
#' @param facets Set to TRUE if you want plot each distribution on a separate facet
#' @return ggplot2 plot object
#' @examples
#' x <- runif(10000)
#' plot_sampling_dist(x, facets = TRUE)
#' plot_sampling_dist(x, n = 1e6, size = c(2, 10, 50, 100))
plot_sampling_dist <- function (pop,   n = 1e5, size = c(2, 5, 30), facets = FALSE) {

  dens <- matrix(ncol = length(size), nrow = n)

  for (i in seq_along(size)) {
    dens[, i] <- replicate(n, mean(sample(pop, size = size[i])))
  }

  colnames(dens) <- paste0("n_", size)
  dens <- as.data.frame(dens)
  dens <- tidyr::gather(dens, size, obs)

  p <- ggplot2::qplot(obs, data = dens, fill = size, color = size,
                      geom = "density", alpha = I(1/3)) +
    ggplot2::ylab("Density") +
    ggplot2::xlab("") +
    ggplot2::theme_bw()
  if (facets) {
    p + ggplot2::facet_grid(size ~ ., scales = "free_y")
  } else {
    p
  }
}

plot_hyper <- function (n_dens = 3) {
  m <- 10
  n <- 10
  x <- 0:10

  dens <- matrix(ncol = n_dens, nrow = length(x))

  for (i in 1:n_dens) {
    dens[, i] <- dhyper(x, m, n, k = n + 2 - i)
  }

  colnames(dens) <- paste0("k_", 1:n_dens)
  dens <- as.data.frame(dens)
  dens <- dplyr::mutate(dens, x = x)
  dens <- tidyr::gather(dens, k, obs, -x)
  ggplot2::qplot(x, obs, data = dens, colour = k, size = k, geom = c("line")) +
    ggplot2::ylab("Probability") +
    ggplot2::xlab("X") +
    ggplot2::theme_bw()
}