Working example for generating multiple plots inside a map2 call

Author

Affiliation

Ronald (Ryy) Glenn Thomas

UCSD

purrr

1 Introduction

Consider the problem of running a data analysis requiring a separate analysis for each of n strata. For example consider an effort to model the relationship between Bill length and Flipper length across three different species of penguins.

We can work with the dataset penguins included in the package palmerpenguins

library(palmerpenguins)

One naive approach is to split the dataset and do three separate analyses:

The R package purrr provides a straightforward method to conduct the analyses with a single command. Assume the set of data tables are contained in a list of dataframes. Also assume the analysis is a simple visualization of a potential linear association between two features,

2 Plots for every variable and each species map inside map see ref 2 below

3 combine plots in a upper triangular grid with correlation coefs

4 Code

library(pacman)
p_load(grid, patchwork, rlang, purrr, palmerpenguins, tidyverse, knitr)

opts_chunk$set(
  warning = FALSE, message = FALSE, echo = FALSE, fig.width = 8,
  fig.height = 9, results = "asis", dev = "pdf"
)
df0 <- sample_n(penguins, 50) |> na.omit()
# nn = 50 ; df1 = sample_n(penguins, 50)
df1 <- split(df0, df0$species)
# df2 = penguins |> group_by(species)


ct <- names(df0)[3:6]
# mm = expand.grid(names(df1[3:6]), names(df1[3:6]))
nn <- t(combn(ct, 2))
colnames(nn) <- letters[1:2]
nn2 <- data.frame(nn) |> cbind(g = "sex")

zz.scatter <- function(data, formula, ...) {
  # function to take a dataframe, and a formula with potentially a '|' group
  #     option and return scatterplot matrix with optional R^2, loess smooth,
  #     or least squares line.
}

plt1 <- function(a, b, g, spc, df_split) {
  out_plot <- df_split |> ggplot(aes(x = .data[[a]], y = .data[[b]])) +
    geom_point(aes(color = .data[[g]]), alpha = .5) +
    geom_smooth() +
    scale_color_manual(values = c("purple", "green", "red")) +
    theme_bw()
  assign(paste0(spc, "_", a, "_", b), value = out_plot, envir = .GlobalEnv)
  return(out_plot)
}


temp <- df1 |> map2(names(df1), function(df_split, spc) {
  nn2 |> pmap(function(a, b, g) {
    plt1(b, a, g, spc, df_split)
  })
})

temp2 = list_flatten(temp)
X <- grid::textGrob("Species 1")
temp2[[2]] = plot_spacer()
temp2[[4]] = X
wrap_plots(temp2, ncol=3, nrow=6) +
  plot_layout(
    guides = "collect",
    axis_titles = "collect"
  ) 
A <- temp[[1]][[1]]
p2 <- temp[[1]][[2]]
p3 <- temp[[1]][[3]]
p4 <- temp[[1]][[4]]
p5 <- temp[[1]][[5]]
p6 <- temp[[1]][[6]]
p7 <- temp[[2]][[1]]
p8 <- temp[[2]][[2]]
p9 <- temp[[2]][[3]]
p10 <- temp[[2]][[4]]
p11 <- temp[[2]][[5]]
p12 <- temp[[2]][[6]]
p13 <- temp[[3]][[1]]
p14 <- temp[[3]][[2]]
p15 <- temp[[3]][[3]]
p16 <- temp[[3]][[4]]
p17 <- temp[[3]][[5]]
p18 <- temp[[3]][[6]]
# names(temp)

layout <- "
X##
ABC
#DE
##F
Y##
GHI
#JK
##L
Z##
MNO
#PQ
##R
"
X <- grid::textGrob("Species 1")
t2 <- grid::textGrob("Species 2")
t3 <- grid::textGrob("Species 3")

out <- wrap_plots(
  X , A, B = p2, C = p3, D = p4, E = p5, F = p6, Y = t2,
  G = p7, H = p8, I = p9, J = p10, K = p11, L = p12, Z = t3,
  M = p13, N = p14, O = p15, P = p16, Q = p17, R = p18,
  design = layout
) +
  plot_layout(
    guides = "collect",
    axis_titles = "collect"
  ) +
  theme(
    legend.position = "bottom",
    legend.direction = "horizontal",
    text = element_text(size = 8)
  )

out

5 References

principal components analysis

Automating exploratory plots with ggplot2 and purrr