如何在R中绘制多层级图表？复现《AppliedPredictiveModeling》指定图表

Got it, let's tackle this plot replication from Applied Predictive Modeling! I'll walk you through how to recreate this 4x4 grid of plots using R (a go-to tool for this book's examples) with ggplot2, which is perfect for this kind of structured visualization.

Step 1: Prepare Your Data #

First, you'll need a dataset that includes these variables:

• x: The x-axis variable for every subplot
• y: The y-axis variable for each curve
• sigma: A variable with 4 distinct values (one per curve in each subplot)
• cost: A variable with 4 distinct values (one per column of subplots)
• epsilon: A variable with 4 distinct values (one per row of subplots)

If you don't have the exact data yet, here's a sample script to generate a matching structure (swap out the y calculation with your actual data logic):

library(ggplot2)
library(dplyr)

# Set seed for reproducible results
set.seed(42)

# Define parameter values (match your actual data's ranges)
sigma_vals <- c(0.1, 0.5, 1.0, 2.0)
cost_vals <- c(1, 5, 10, 20)
epsilon_vals <- c(0.01, 0.05, 0.1, 0.2)

# Create all combinations of parameters + x values
plot_data <- expand.grid(
  x = seq(0, 10, length.out = 100),
  sigma = sigma_vals,
  cost = cost_vals,
  epsilon = epsilon_vals
)

# Generate y values (replace this with your real data's formula)
plot_data <- plot_data %>%
  mutate(y = sin(x) + sigma * rnorm(nrow(.)) + epsilon * x + (cost / 10))

Step 2: Build the Faceted Plot #

This code will generate the 16-subplot grid exactly as you described:

ggplot(plot_data, aes(x = x, y = y, color = factor(sigma))) +
  # Draw 4 distinct curves per subplot (one per sigma)
  geom_line(linewidth = 1) +
  # Create 4x4 grid: rows = epsilon values, columns = cost values
  facet_grid(epsilon ~ cost, labeller = label_both) +
  # Customize legend and axis labels
  scale_color_discrete(name = "Sigma") +
  labs(x = "X Axis", y = "Y Axis") +
  # Clean up theme for readability
  theme_bw() +
  theme(
    strip.background = element_rect(fill = "#f0f0f0"),
    legend.position = "bottom",
    legend.box = "horizontal"
  )

Quick Notes to Match Your Target Plot: #

• facet_grid(epsilon ~ cost) ensures rows map to epsilon values and columns map to cost values, creating the 4x4 layout.
• color = factor(sigma) assigns a unique color to each sigma value, so you get 4 distinct lines in every subplot.
• Swap out the simulated y calculation with your actual dataset's logic to replicate the original curves perfectly.

内容的提问来源于stack exchange，提问作者Marcus Grass