Facets, Scales, Labels and Themes

From exploration to communication

Exploratory plots are quick; plots for reports need care. Facets split one plot into small multiples by a variable: facet_wrap(~ city) or facet_grid(rows ~ cols), often clearer than crowding many colours into one panel. Scales control how data maps to visuals: scale_y_continuous(labels = scales::label_comma()) or scales::label_currency(prefix = "₹") for readable numbers, scale_colour_brewer() or scale_colour_viridis_d() for colour-blind-friendly palettes, and log scales for skewed data. labs() sets the title, subtitle, axis labels, caption (cite your data source) and legend titles. Themes change the look: theme_minimal(), theme_bw(), and theme() adjustments for font sizes, legend position and gridlines; you can set a default with theme_set(). Extensions such as patchwork combine plots, ggrepel avoids overlapping labels, and plotly (ggplotly()) adds interactivity. Good charts start the y-axis at zero for bars, avoid 3D effects, and label directly where possible.

A polished faceted chart

Readable currency labels, a colour-blind-friendly palette and a clean theme.

library(ggplot2)
library(scales)

monthly <- sales |>
  dplyr::mutate(month = lubridate::floor_date(order_date, "month")) |>
  dplyr::summarise(revenue = sum(amount_inr), .by = c(city, month))

ggplot(monthly, aes(x = month, y = revenue, colour = city)) +
  geom_line(linewidth = 1) +
  geom_point() +
  facet_wrap(~ city, ncol = 2) +
  scale_y_continuous(labels = label_comma(prefix = "Rs ")) +
  scale_colour_viridis_d(guide = "none") +        # facets already identify cities
  labs(
    title = "Monthly revenue by city",
    subtitle = "Financial year 2026-27",
    x = NULL, y = "Revenue",
    caption = "Source: internal orders database"
  ) +
  theme_minimal(base_size = 12)

Tidying a room before guests arrive

An exploratory plot is a room you live in: functional but messy. A report plot is the same room tidied for guests: labels on the doors, colours that match, and nothing distracting on the floor.

Quick check: What does facet_wrap(~ city) do?

• Changes the colour palette
• Sorts the data
• Saves the plot
• Creates small multiples: one panel per city