Faceting and Multi-Panel Layouts with ggplot2

Setup

library(readr)
library(ggplot2)
library(dplyr)
library(patchwork)

# Load data
data_source <- "https://raw.githubusercontent.com/simoneSantoni/data-viz-smm635/refs/heads/master/data/googleplaystore.csv"
apps <- read_csv(data_source)

# Clean data
apps <- apps %>%
  mutate(
    Reviews = as.numeric(Reviews),
    Rating = as.numeric(Rating)
  ) %>%
  filter(Type %in% c("Free", "Paid"))

This document demonstrates advanced layout techniques in ggplot2:

• Faceting: Built-in ggplot2 features for creating small multiples
• Patchwork: A powerful extension for combining independent plots with flexible layouts

Faceting in ggplot2

Faceting creates multiple panels based on categorical variables, allowing comparison across subgroups. ggplot2 provides two main faceting functions:

facet_wrap()

facet_wrap() wraps a 1D sequence of panels into 2D, useful for faceting by a single variable.

Basic Faceting

apps %>%
  ggplot(aes(x = Rating)) +
  geom_histogram(bins = 30, fill = "steelblue", alpha = 0.7) +
  facet_wrap(~Type) +
  theme_minimal() +
  labs(title = "Rating Distribution by App Type",
       x = "Rating", y = "Count")

Custom Layout

apps %>%
  filter(`Content Rating` %in% c("Everyone", "Teen", "Mature 17+", "Everyone 10+")) %>%
  ggplot(aes(x = Rating)) +
  geom_density(fill = "coral", alpha = 0.6) +
  facet_wrap(~`Content Rating`, ncol = 2) +
  theme_minimal() +
  labs(title = "Rating Density by Content Rating",
       x = "Rating", y = "Density")

Free Scales

apps %>%
  filter(`Content Rating` %in% c("Everyone", "Teen", "Mature 17+")) %>%
  ggplot(aes(x = Reviews)) +
  geom_histogram(bins = 20, fill = "darkgreen", alpha = 0.7) +
  facet_wrap(~`Content Rating`, scales = "free") +
  scale_x_log10() +
  theme_minimal() +
  labs(title = "Review Distribution with Independent Scales",
       x = "Reviews (log scale)", y = "Count")

facet_grid()

facet_grid() forms a matrix of panels defined by row and column faceting variables. Perfect for examining interactions between two categorical variables.

Grid by Two Variables

apps %>%
  filter(`Content Rating` %in% c("Everyone", "Teen")) %>%
  ggplot(aes(x = Reviews, y = Rating)) +
  geom_point(alpha = 0.3, color = "purple") +
  geom_smooth(method = "lm", color = "red", se = FALSE) +
  facet_grid(Type ~ `Content Rating`) +
  scale_x_log10() +
  theme_minimal() +
  labs(title = "Reviews vs Rating: Type � Content Rating",
       x = "Reviews (log scale)", y = "Rating")

Row Facets Only

apps %>%
  ggplot(aes(x = Rating, fill = Type)) +
  geom_density(alpha = 0.5) +
  facet_grid(Type ~ .) +
  theme_minimal() +
  scale_fill_brewer(palette = "Set1") +
  labs(title = "Rating Density by App Type (Row Facets)",
       x = "Rating", y = "Density")

Column Facets Only

apps %>%
  ggplot(aes(x = Rating, fill = Type)) +
  geom_histogram(bins = 30, alpha = 0.7, position = "identity") +
  facet_grid(. ~ Type) +
  theme_minimal() +
  scale_fill_brewer(palette = "Set2") +
  labs(title = "Rating Distribution by App Type (Column Facets)",
       x = "Rating", y = "Count")

Patchwork: Flexible Multi-Panel Layouts

The patchwork package extends ggplot2’s capabilities by allowing you to combine independent plots with intuitive operators.

Basic Patchwork Operators

Side-by-Side (+)

p1 <- apps %>%
  ggplot(aes(x = Type, fill = Type)) +
  geom_bar() +
  theme_minimal() +
  scale_fill_brewer(palette = "Pastel1") +
  labs(title = "App Count by Type", x = "", y = "Count")

p2 <- apps %>%
  ggplot(aes(x = Type, y = Rating, fill = Type)) +
  geom_boxplot() +
  theme_minimal() +
  scale_fill_brewer(palette = "Pastel1") +
  labs(title = "Rating Distribution", x = "", y = "Rating")

p1 + p2

Stacked (/)

p3 <- apps %>%
  filter(Reviews > 0) %>%
  ggplot(aes(x = Reviews)) +
  geom_histogram(bins = 50, fill = "skyblue") +
  scale_x_log10() +
  theme_minimal() +
  labs(title = "Review Count Distribution", x = "Reviews (log scale)", y = "Count")

p4 <- apps %>%
  ggplot(aes(x = Rating)) +
  geom_histogram(bins = 30, fill = "salmon") +
  theme_minimal() +
  labs(title = "Rating Distribution", x = "Rating", y = "Count")

p3 / p4

Complex Layouts

p5 <- apps %>%
  ggplot(aes(x = `Content Rating`, fill = Type)) +
  geom_bar(position = "dodge") +
  theme_minimal() +
  theme(axis.text.x = element_text(angle = 45, hjust = 1)) +
  scale_fill_brewer(palette = "Set3") +
  labs(title = "Apps by Content Rating & Type", x = "", y = "Count")

(p1 + p2) / p5

Advanced Patchwork Features

Plot Annotations

combined <- p1 + p2 + p3 + p4

combined +
  plot_annotation(
    title = 'Google Play Store Apps Analysis',
    subtitle = 'Distribution of key metrics across app types',
    tag_levels = 'A',
    theme = theme(plot.title = element_text(size = 16, face = "bold"))
  )

Layout Control

p1 + p2 + p3 + p4 +
  plot_layout(ncol = 2, heights = c(1, 2))

Nested Layouts

p6 <- apps %>%
  filter(Reviews > 0, !is.na(Rating)) %>%
  ggplot(aes(x = Reviews, y = Rating, color = Type)) +
  geom_point(alpha = 0.4) +
  geom_smooth(method = "lm", se = FALSE) +
  scale_x_log10() +
  scale_color_brewer(palette = "Set1") +
  theme_minimal() +
  labs(title = "Reviews vs Rating", x = "Reviews (log)", y = "Rating")

p7 <- apps %>%
  count(Type, `Content Rating`) %>%
  filter(`Content Rating` %in% c("Everyone", "Teen", "Mature 17+")) %>%
  ggplot(aes(x = `Content Rating`, y = n, fill = Type)) +
  geom_col(position = "dodge") +
  theme_minimal() +
  scale_fill_brewer(palette = "Set1") +
  labs(title = "Distribution by Content Rating", x = "", y = "Count")

(p6 | (p1 / p2)) + plot_layout(widths = c(2, 1))

Collecting Guides

When plots share legends, patchwork can collect them into a single legend:

p8 <- apps %>%
  filter(Reviews > 0) %>%
  ggplot(aes(x = Reviews, fill = Type)) +
  geom_histogram(bins = 30, alpha = 0.7) +
  scale_x_log10() +
  scale_fill_brewer(palette = "Dark2") +
  theme_minimal() +
  labs(title = "Review Distribution", x = "Reviews (log)", y = "Count")

p9 <- apps %>%
  filter(!is.na(Rating)) %>%
  ggplot(aes(x = Rating, fill = Type)) +
  geom_density(alpha = 0.6) +
  scale_fill_brewer(palette = "Dark2") +
  theme_minimal() +
  labs(title = "Rating Density", x = "Rating", y = "Density")

p8 + p9 + plot_layout(guides = "collect")

Faceting vs Patchwork: When to Use Which?

Feature	facet_wrap / facet_grid	patchwork
Use case	Same plot type, same variables, different subgroups	Different plot types, different variables, or different data
Data source	Single data frame	Multiple data frames or sources
Aesthetics	Shared across all panels	Independent per plot
Scales	Can be fixed or freed	Fully independent
Layout	Regular grid	Flexible, asymmetric layouts
Legends	Automatically unified	Can be collected or kept separate
Annotations	Standard ggplot2 labels	Advanced with plot_annotation()

TipBest Practices

• Use faceting when exploring how a single relationship varies across categories
• Use patchwork when combining conceptually different visualizations into a dashboard
• Combine both: Facet individual plots, then use patchwork to arrange them

Example: Complete Analysis Dashboard

# Overview panel
overview <- apps %>%
  count(Type, `Content Rating`) %>%
  filter(`Content Rating` %in% c("Everyone", "Teen", "Mature 17+", "Everyone 10+")) %>%
  ggplot(aes(x = `Content Rating`, y = n, fill = Type)) +
  geom_col(position = "dodge") +
  theme_minimal() +
  scale_fill_brewer(palette = "Set1") +
  theme(axis.text.x = element_text(angle = 45, hjust = 1)) +
  labs(title = "App Distribution", x = "", y = "Count")

# Rating analysis with facets
rating_facet <- apps %>%
  filter(`Content Rating` %in% c("Everyone", "Teen")) %>%
  ggplot(aes(x = Rating, fill = Type)) +
  geom_density(alpha = 0.6) +
  facet_wrap(~`Content Rating`) +
  scale_fill_brewer(palette = "Set1") +
  theme_minimal() +
  labs(title = "Rating Density by Content Rating", x = "Rating", y = "Density")

# Reviews analysis
reviews_scatter <- apps %>%
  filter(Reviews > 0, !is.na(Rating)) %>%
  ggplot(aes(x = Reviews, y = Rating, color = Type)) +
  geom_point(alpha = 0.3, size = 0.8) +
  geom_smooth(method = "lm", se = FALSE, size = 1) +
  scale_x_log10() +
  scale_color_brewer(palette = "Set1") +
  theme_minimal() +
  labs(title = "Reviews vs Rating", x = "Reviews (log scale)", y = "Rating")

# Combine with patchwork
(overview | rating_facet) / reviews_scatter +
  plot_layout(heights = c(1, 1.5), guides = "collect") +
  plot_annotation(
    title = 'Google Play Store Apps: Comprehensive Analysis',
    subtitle = 'Combining faceting and patchwork for effective data storytelling',
    tag_levels = 'A',
    theme = theme(
      plot.title = element_text(size = 18, face = "bold"),
      plot.subtitle = element_text(size = 12)
    )
  ) &
  theme(legend.position = "bottom")

Summary

This document demonstrated:

1. facet_wrap(): For wrapping panels in a ribbon
2. facet_grid(): For creating rectangular grids
3. Patchwork operators: + (side-by-side), / (stacked), | (compose)
4. plot_layout(): Fine control over dimensions and arrangement
5. plot_annotation(): Adding titles, tags, and themes
6. Collecting guides: Unifying legends across plots

ImportantFurther Learning

• Patchwork GitHub
• Patchwork documentation
• ggplot2 book on faceting