ggplot2 Facets Exercises in R: 20 Practice Problems

Exercise 1.1: Split a scatterplot into one panel per car class

Task: A used-car analyst is comparing how engine displacement relates to highway mpg for each vehicle category in the built-in mpg dataset. Build a scatterplot of hwy against displ and use facet_wrap() to split it into one panel per class. Save the plot to ex_1_1.

Expected result:

# A faceted scatterplot: x = displ, y = hwy, points in default black,
# one panel per class (compact, midsize, minivan, pickup, subcompact, suv, 2seater).
# Default 3-column layout with strip labels at the top of each panel.

Difficulty: Beginner

RYour turn

ex_1_1 <- # your code here ex_1_1

Click to reveal solution

RSolution

ex_1_1 <- ggplot(mpg, aes(displ, hwy)) + geom_point() + facet_wrap(~ class) ex_1_1

Explanation: facet_wrap() takes a one-sided formula ~ class and generates one panel per level of class, packing them into a rough grid. ggplot picks a default ncol from the panel count and the plot aspect ratio. The formula form ~ class is shorthand for vars(class), and the vars() style is preferred in newer code because it cooperates better with tidy-eval helpers when you want to facet programmatically.

Exercise 1.2: Force a two-row layout with nrow

Task: Reusing the mpg scatterplot of hwy against displ, rearrange the seven class panels into exactly two rows so they fit cleanly inside a wide report column without losing readability. Save the plot to ex_1_2.

Expected result:

# Same scatter as 1.1, but panels arranged in exactly 2 rows by 4 columns.
# The eighth cell stays empty because there are only 7 classes.

Difficulty: Beginner

RYour turn

ex_1_2 <- # your code here ex_1_2

Click to reveal solution

RSolution

ex_1_2 <- ggplot(mpg, aes(displ, hwy)) + geom_point() + facet_wrap(~ class, nrow = 2) ex_1_2

Explanation: Set nrow = 2 and ggplot picks the column count automatically. Pin one or the other (nrow OR ncol), not both: if you set both and they undercount the levels, ggplot silently drops panels. The cleaner pattern is to fix the dimension that matches your page geometry and let ggplot pick the other.

Exercise 1.3: Let each panel pick its own y-axis range

Task: A retail analyst inspecting the txhousing dataset wants a line chart of monthly median price by date for the cities Austin, Dallas, Houston, San Antonio, and Fort Worth, with each city on its own y-axis range so small-market detail does not get crushed by Houston. Save the plot to ex_1_3.

Expected result:

# 5 panels of median price line over time, one per city.
# Each panel uses its own y-axis range (Austin roughly 80k to 300k,
# Dallas roughly 115k to 220k, etc). The x-axis date is shared.

Difficulty: Intermediate

RYour turn

big_5 <- txhousing |> filter(city %in% c("Austin", "Dallas", "Houston", "San Antonio", "Fort Worth")) ex_1_3 <- # your code here ex_1_3

Click to reveal solution

RSolution

big_5 <- txhousing |> filter(city %in% c("Austin", "Dallas", "Houston", "San Antonio", "Fort Worth")) ex_1_3 <- ggplot(big_5, aes(date, median)) + geom_line() + facet_wrap(~ city, scales = "free_y") ex_1_3

Explanation: scales = "free_y" lets each panel rescale its y-axis independently while keeping the x-axis common. It is the right call when panel magnitudes differ but trend shape is what matters. scales = "free" frees both axes; "free_x" frees only the x. For strict cross-city comparison of absolute price, leave it at the default "fixed" so the eye reads magnitude correctly.

Exercise 1.4: Rename facet strip labels without renaming the column

Task: The previous Texas housing chart shows "Austin", "Dallas", and the rest on its strip labels, but the executive deck needs "Austin TX", "Dallas TX", and so on. Rebuild the facet plot using as_labeller() so the source column stays unchanged but the strip text appends " TX" to each city. Save the plot to ex_1_4.

Expected result:

# Same 5-panel layout as 1.3 but strip labels read
# "Austin TX", "Dallas TX", "Fort Worth TX", "Houston TX", "San Antonio TX".

Difficulty: Intermediate

RYour turn

ex_1_4 <- # your code here ex_1_4

Click to reveal solution

RSolution

city_labeller <- as_labeller(function(x) paste(x, "TX")) ex_1_4 <- ggplot(big_5, aes(date, median)) + geom_line() + facet_wrap(~ city, scales = "free_y", labeller = city_labeller) ex_1_4

Explanation: as_labeller() wraps any function so ggplot treats it as a panel-label transformer. A named character vector works too: as_labeller(c(Austin = "Austin TX", ...)) for one-off cases. The advantage over editing the source column is that joins, sorts, and downstream summaries still see clean values like "Austin"; only the display layer changes.

Exercise 2.1: Cross-classify points with facet_grid

Task: A used-car reviewer wants to see hwy vs displ from mpg cross-tabulated by drivetrain (drv) on rows and number of cylinders (cyl) on columns. Build the scatter and apply facet_grid() with drv on the y-direction and cyl on the x-direction. Save the plot to ex_2_1.

Expected result:

# Grid of 3 rows (drv = 4, f, r) by 4 cols (cyl = 4, 5, 6, 8) for up to 12 panels.
# Some panels stay empty (e.g. rear-drive 4-cylinder); shared x and y axes across all.

Difficulty: Intermediate

RYour turn

ex_2_1 <- # your code here ex_2_1

Click to reveal solution

RSolution

ex_2_1 <- ggplot(mpg, aes(displ, hwy)) + geom_point() + facet_grid(drv ~ cyl) ex_2_1

Explanation: facet_grid() lays panels out on a strict rows-by-columns matrix and shows empty cells for missing combinations, which is useful for spotting sparsity (no rear-drive 4-cylinder, no 5-cyl rear-drive) at a glance. Compare with facet_wrap(), which packs only the present combinations and loses the cross-tab structure. Use grid when the row and column dimensions carry meaning; wrap when they do not.

Exercise 2.2: Stack panels in a single column with the dot placeholder

Task: The same used-car scatter of hwy against displ from mpg needs to be split into three stacked panels, one per drivetrain (drv), all sharing the x-axis so vertical comparison is straightforward. Use facet_grid() with the . placeholder for columns. Save the plot to ex_2_2.

Expected result:

# 3 stacked panels (drv = 4, then f, then r), each spanning the full plot width.
# Shared x = displ, shared y = hwy. Strip labels sit on the right edge.

Difficulty: Intermediate

RYour turn

ex_2_2 <- # your code here ex_2_2

Click to reveal solution

RSolution

ex_2_2 <- ggplot(mpg, aes(displ, hwy)) + geom_point() + facet_grid(drv ~ .) ex_2_2

Explanation: drv ~ . reads as "facet rows by drv, no column variable." Mirror image . ~ drv gives three side-by-side panels instead. Stacking is the cleanest way to compare distributions vertically because the x-axis is locked and the eye scans cleanly down. facet_wrap(~ drv, ncol = 1) does almost the same thing, but facet_grid keeps the strip on a fixed side and never reflows.

Exercise 2.3: Show marginal totals with facet_grid margins

Task: Continuing with the mpg data, build a facet_grid(drv ~ cyl) scatter of hwy vs displ and add marginal panels along the right side and bottom that summarize across the other factor. The marginals make the per-row, per-column, and overall trends visible in one figure. Save the plot to ex_2_3.

Expected result:

# Same 3x4 grid as 2.1 plus an extra "(all)" row and "(all)" column
# containing the rolled-up scatter across the marginalized dimension.
# Net layout: 4 rows by 5 cols, last row and last col labeled "(all)".

Difficulty: Advanced

RYour turn

ex_2_3 <- # your code here ex_2_3

Click to reveal solution

RSolution

ex_2_3 <- ggplot(mpg, aes(displ, hwy)) + geom_point() + facet_grid(drv ~ cyl, margins = TRUE) ex_2_3

Explanation: margins = TRUE adds an extra row, an extra column, and a corner panel that aggregate across the other dimension. You can also pass a vector like margins = "drv" to add only the row margin. Marginals are essential when stakeholders want both the slice and the overall picture in one chart. Watch out: every observation now appears in multiple panels (one cell, one row-margin, one col-margin, one grand total), so do not compute counts from the rendered figure.

Exercise 2.4: Free the y-axis per row in a facet_grid

Task: A trial analyst studying ChickWeight wants to look at weight trajectories on a Time x-axis split into one row per Diet (1, 2, 3, 4), with each row allowed its own y-axis range because diet 3 and diet 4 trend higher than the other two. Use facet_grid() with the appropriate scales argument. Save the plot to ex_2_4.

Expected result:

# 4 stacked panels (Diet 1 to 4) of weight vs Time, one line per Chick.
# Each row has its own y-axis range; the x-axis Time is shared.

Difficulty: Intermediate

RYour turn

ex_2_4 <- # your code here ex_2_4

Click to reveal solution

RSolution

ex_2_4 <- ggplot(ChickWeight, aes(Time, weight, group = Chick)) + geom_line(alpha = 0.5) + facet_grid(Diet ~ ., scales = "free_y") ex_2_4

Explanation: In facet_grid(), scales = "free_y" frees the y per row (not per cell), and "free_x" frees the x per column. This is because grid preserves the cross-tab structure: every panel in a row shares a y by design, every panel in a column shares an x. Only facet_wrap() lets every individual panel float on both axes via scales = "free".

Exercise 3.1: Allocate panel height proportional to bar count

Task: Filter mpg to four manufacturers (audi, chevrolet, honda, toyota), count vehicles by class within each, and build a horizontal bar chart faceted by manufacturer in stacked rows. Use space = "free_y" so panels with fewer classes occupy less vertical space than panels with more. Save the plot to ex_3_1.

Expected result:

# 4 horizontally-stacked panels (audi, chevrolet, honda, toyota).
# Each panel: y = class, x = n. Panel height varies: honda (2 classes) shortest,
# chevrolet (5 classes) tallest. Default geom_col() blue-grey bars of equal thickness.

Difficulty: Advanced

RYour turn

mfr_counts <- mpg |> filter(manufacturer %in% c("audi", "chevrolet", "honda", "toyota")) |> count(manufacturer, class) ex_3_1 <- # your code here ex_3_1

Click to reveal solution

RSolution

mfr_counts <- mpg |> filter(manufacturer %in% c("audi", "chevrolet", "honda", "toyota")) |> count(manufacturer, class) ex_3_1 <- ggplot(mfr_counts, aes(n, class)) + geom_col() + facet_grid(manufacturer ~ ., scales = "free_y", space = "free_y") ex_3_1

Explanation: scales = "free_y" lets each panel show only the categories that actually exist in that subset (otherwise honda would display empty bars for "pickup", "minivan", and the rest). space = "free_y" then scales the row heights with the category count, so every bar stays the same physical thickness across panels. Without space, every panel gets equal height and the bars in shorter panels balloon while bars in taller panels get squeezed thin.

Exercise 3.2: Fix the lower y-bound but free the upper bound per panel

Task: A trial analyst showing the ChickWeight weight curves faceted by Diet wants each panel's y-axis to start at zero but stretch upward to whatever value its own data needs. Build the plot using a layered blank data frame that injects a y = 0 anchor into every panel, then apply scales = "free_y". Save the plot to ex_3_2.

Expected result:

# 4 panels (Diet 1 to 4) of weight vs Time, one line per Chick.
# Each y-axis starts cleanly at 0; the upper bound varies per panel
# (Diet 1 reaches roughly 300, Diet 3 reaches roughly 370).

Difficulty: Advanced

RYour turn

ex_3_2 <- # your code here ex_3_2

Click to reveal solution

RSolution

zero_anchor <- data.frame( Diet = factor(1:4), Time = 0, weight = 0 ) ex_3_2 <- ggplot(ChickWeight, aes(Time, weight, group = Chick)) + geom_line(alpha = 0.5) + geom_blank(data = zero_anchor, aes(Time, weight, group = NULL)) + facet_wrap(~ Diet, scales = "free_y") ex_3_2

Explanation: coord_cartesian(ylim = c(0, NA)) would clip every panel to the same range and defeat scales = "free_y". geom_blank() is the idiomatic workaround: it injects an invisible point in each panel that anchors the lower bound at 0 without forcing an upper bound. The same trick works for pinning an upper limit or guaranteeing a specific tick mark appears in every panel.

Exercise 3.3: Reorder facet panels by a summary statistic

Task: A reviewer wants mpg faceted by class but with panels arranged by descending median hwy so the most fuel-efficient class appears in the top-left, not in alphabetical order. Reorder class into a factor using reorder() before plotting, then use facet_wrap(). Save the plot to ex_3_3.

Expected result:

# 7 panels in this order: subcompact, compact, midsize, 2seater, minivan, suv, pickup
# (descending median hwy). Each panel is the usual hwy-vs-displ scatter.

Difficulty: Intermediate

RYour turn

ex_3_3 <- # your code here ex_3_3

Click to reveal solution

RSolution

mpg_ord <- mpg |> mutate(class = reorder(class, -hwy, FUN = median)) ex_3_3 <- ggplot(mpg_ord, aes(displ, hwy)) + geom_point() + facet_wrap(~ class) ex_3_3

Explanation: Facet panel order follows the factor-level order of the faceting variable. Base R's reorder() re-levels by a summary statistic (median, mean, custom function), and the negative sign on -hwy flips the order to descending. The forcats equivalent is fct_reorder(class, hwy, .fun = median, .desc = TRUE), which reads more clearly when the rest of the pipeline uses tidyverse helpers.

Exercise 4.1: Style facet strips with a custom theme

Task: The Texas housing chart from earlier is being repurposed for a brand-aligned PDF. Restyle each panel's strip background to a dark slate fill with bold white text at 11 point, using theme() elements. Apply the change to the five-city price-over-time facet plot. Save the plot to ex_4_1.

Expected result:

# 5 facet panels of median price vs date.
# Strip backgrounds rendered in slate (#2C3E50) with white bold 11pt text labels.
# Plot body (lines, axes) unchanged from 1.3.

Difficulty: Intermediate

RYour turn

ex_4_1 <- # your code here ex_4_1

Click to reveal solution

RSolution

ex_4_1 <- ggplot(big_5, aes(date, median)) + geom_line() + facet_wrap(~ city, scales = "free_y") + theme( strip.background = element_rect(fill = "#2C3E50", colour = NA), strip.text = element_text(colour = "white", face = "bold", size = 11) ) ex_4_1

Explanation: strip.background controls the rectangle behind each strip label; strip.text controls the text itself. Both accept the same element_rect() and element_text() arguments familiar from any ggplot theme tweak. For top-only or side-only customization, use the suffixed variants strip.background.x, strip.background.y, strip.text.x, strip.text.y instead.

Exercise 4.2: Wrap long facet labels onto multiple lines

Task: Build a price-vs-carat scatter of diamonds, faceted by a derived clarity_long column that uses full descriptive grade names like "Very Very Slightly Included 1". Apply label_wrap_gen() so labels wider than 18 characters wrap onto two lines instead of overflowing the strip. Save the plot to ex_4_2.

Expected result:

# 8 panels of price vs carat scatter. Strip labels show the full grade names
# wrapped at roughly 18 chars (e.g. "Very Slightly\nIncluded 2"), no overflow.

Difficulty: Intermediate

RYour turn

diamonds_long <- diamonds |> mutate(clarity_long = recode(clarity, I1 = "Included Grade 1", SI2 = "Slightly Included 2", SI1 = "Slightly Included 1", VS2 = "Very Slightly Included 2", VS1 = "Very Slightly Included 1", VVS2 = "Very Very Slightly Included 2", VVS1 = "Very Very Slightly Included 1", IF = "Internally Flawless" )) ex_4_2 <- # your code here ex_4_2

Click to reveal solution

RSolution

diamonds_long <- diamonds |> mutate(clarity_long = recode(clarity, I1 = "Included Grade 1", SI2 = "Slightly Included 2", SI1 = "Slightly Included 1", VS2 = "Very Slightly Included 2", VS1 = "Very Slightly Included 1", VVS2 = "Very Very Slightly Included 2", VVS1 = "Very Very Slightly Included 1", IF = "Internally Flawless" )) ex_4_2 <- ggplot(diamonds_long, aes(carat, price)) + geom_point(alpha = 0.05) + facet_wrap(~ clarity_long, labeller = label_wrap_gen(width = 18)) ex_4_2

Explanation: label_wrap_gen(width = N) is a labeller factory that returns a function ggplot can call on every panel label. It breaks at word boundaries near width and inserts newlines, which the strip renders as multi-line text. The alternative as_labeller(function(x) stringr::str_wrap(x, 18)) gives you finer control if the boundaries matter; label_value is the default when no wrapping is needed at all.

Exercise 4.3: Show row and column variable names together with label_both

Task: The car-shopper wants the mpg data faceted by drv on rows and cyl on columns, with each strip label including the variable name as well as the level (e.g. "drv: 4" and "cyl: 6") instead of bare values, which makes the figure self-explanatory in a slide. Save the plot to ex_4_3.

Expected result:

# 3 by 4 grid of hwy-vs-displ scatter panels.
# Strips read "drv: 4" / "drv: f" / "drv: r" on the right
# and "cyl: 4" / "cyl: 5" / "cyl: 6" / "cyl: 8" on top.

Difficulty: Intermediate

RYour turn

ex_4_3 <- # your code here ex_4_3

Click to reveal solution

RSolution

ex_4_3 <- ggplot(mpg, aes(displ, hwy)) + geom_point() + facet_grid(drv ~ cyl, labeller = label_both) ex_4_3

Explanation: label_both is a ready-made labeller that prefixes each level with its variable name. It is a quick win for diagnostic plots where panel context is not obvious. The alternatives are label_value (default; just the level), label_parsed (treats labels as plotmath expressions), and label_context (mimics R's print for the variable's class). Combine labellers per facet axis with labeller(drv = label_both, cyl = label_value).

Exercise 4.4: Move facet strips to the bottom of each panel

Task: A poster designer wants the strip labels on a facet_wrap chart to sit BELOW each panel instead of above, freeing vertical space at the top for a large headline title. Build the mpg scatter faceted by class and place strips at the bottom using strip.position. Save the plot to ex_4_4.

Expected result:

# Same 7-panel hwy-vs-displ scatter as 1.1, but the class strip labels
# ("compact", "midsize", etc.) appear directly underneath each panel rather than above.

Difficulty: Beginner

RYour turn

ex_4_4 <- # your code here ex_4_4

Click to reveal solution

RSolution

ex_4_4 <- ggplot(mpg, aes(displ, hwy)) + geom_point() + facet_wrap(~ class, strip.position = "bottom") ex_4_4

Explanation: strip.position in facet_wrap() accepts "top" (default), "bottom", "left", or "right". For facet_grid(), use the switch argument instead: switch = "x" moves column strips to the bottom, switch = "y" moves row strips to the left, switch = "both" moves both. Bottom strips pair well with a top-anchored title that needs to dominate the space.

Exercise 5.1: Small multiples for a long-format time series

Task: Reshape the built-in economics dataset into long format with one row per (date, metric) combination, then plot every metric over date as a small-multiples line chart using facet_wrap() with scales = "free_y". The result is a quick dashboard of all five macro indicators. Save the plot to ex_5_1.

Expected result:

# 5 line panels (pce, pop, psavert, uempmed, unemploy) of value over date.
# Each y-axis has its own scale; the x-axis (date 1967 to 2015) is shared.
# Default 3-column layout.

Difficulty: Intermediate

RYour turn

ex_5_1 <- # your code here ex_5_1

Click to reveal solution

RSolution

econ_long <- economics |> pivot_longer(-date, names_to = "metric", values_to = "value") ex_5_1 <- ggplot(econ_long, aes(date, value)) + geom_line() + facet_wrap(~ metric, scales = "free_y") ex_5_1

Explanation: Pivoting to long format is the standard prep for small multiples in ggplot: one row per "thing being plotted" lets facet_wrap() partition the data cleanly. scales = "free_y" is non-negotiable here because the metrics span very different magnitudes (population in hundreds of millions, savings rate in single-digit percent). The retired predecessor gather() still runs but pivot_longer() is the current idiom.

Exercise 5.2: Facet a continuous variable by binning

Task: Bin the carat column of diamonds into four quartile groups using ntile(), then build a price-vs-depth scatter faceted by the resulting bin so each panel shows a roughly equal-sized slice of the carat distribution. Save the plot to ex_5_2.

Expected result:

# 4 panels (carat quartile Q1, Q2, Q3, Q4) of price vs depth scatter.
# Each panel holds roughly 25% of rows (~13,485 diamonds).

Difficulty: Intermediate

RYour turn

ex_5_2 <- # your code here ex_5_2

Click to reveal solution

RSolution

diamonds_q <- diamonds |> mutate(carat_q = paste0("Q", ntile(carat, 4))) ex_5_2 <- ggplot(diamonds_q, aes(depth, price)) + geom_point(alpha = 0.05) + facet_wrap(~ carat_q) ex_5_2

Explanation: Facets need a discrete variable, so binning a continuous one is the standard bridge. ntile() from dplyr produces n equal-frequency bins; cut() from base produces equal-width bins; cut_number() from ggplot2 wraps the equal-frequency idea with auto-generated interval labels. Equal-frequency bins are best when distribution shape is the focus; equal-width when absolute value ranges are.

Exercise 5.3: Highlight one panel by greying the rest

Task: Build a mpg scatter of hwy against displ faceted by class, but overlay the FULL dataset in light grey behind each panel so the focal class points stand out against the full-distribution backdrop. Use a layered-data trick: one geom with the global data, one with the per-panel data. Save the plot to ex_5_3.

Expected result:

# 7 panels. Each panel shows ALL ~234 mpg points in light grey
# with the panel's own class points overlaid in black.
# The grey "ghost" reveals how each class sits inside the overall distribution.

Difficulty: Advanced

RYour turn

ex_5_3 <- # your code here ex_5_3

Click to reveal solution

RSolution

mpg_bg <- mpg |> select(-class) ex_5_3 <- ggplot(mpg, aes(displ, hwy)) + geom_point(data = mpg_bg, colour = "grey80") + geom_point() + facet_wrap(~ class) ex_5_3

Explanation: Dropping the faceting column from the background data frame is the trick: with class absent, ggplot has nothing to filter on per panel and replicates those points across every panel instead. Layer the panel-specific data on top so the focal points stay readable. This pattern beats the gghighlight package for simple cases because there are no extra dependencies and the draw order avoids antialiasing artifacts.

Exercise 5.4: Add a per-panel summary statistic as a label

Task: For each class panel in a mpg scatter of hwy vs displ, annotate the top-left corner with that panel's mean highway mpg using geom_text() driven by a per-class summary data frame. The label should anchor regardless of panel scale. Save the plot to ex_5_4.

Expected result:

# 7 facet panels of hwy vs displ. Top-left of each panel has black text
# like "mean hwy = 28.3" for compact and "mean hwy = 16.9" for pickup.

Difficulty: Intermediate

RYour turn

ex_5_4 <- # your code here ex_5_4

Click to reveal solution

RSolution

class_means <- mpg |> group_by(class) |> summarise(mean_hwy = mean(hwy), .groups = "drop") |> mutate(label = sprintf("mean hwy = %.1f", mean_hwy)) ex_5_4 <- ggplot(mpg, aes(displ, hwy)) + geom_point() + geom_text( data = class_means, aes(x = -Inf, y = Inf, label = label), hjust = -0.1, vjust = 1.2, inherit.aes = FALSE ) + facet_wrap(~ class) ex_5_4

Explanation: Setting x = -Inf and y = Inf plus hjust / vjust anchors the text to the top-left corner regardless of panel scale, so the label survives scales = "free". inherit.aes = FALSE decouples the text layer from the global aesthetics so ggplot does not complain about missing displ or hwy columns in class_means. The summary data frame MUST carry the same faceting column so ggplot can route each label to the right panel.

Exercise 5.5: Use facet_grid with two row variables via vars()

Task: A car-shopper wants mpg scatter panels laid out with both drv and year combined as the row dimension (so each unique drv-year pair gets its own row) and cyl on the columns. Use vars() to pass multiple faceting variables to facet_grid(). Save the plot to ex_5_5.

Expected result:

# 6 rows (drv x year combos: 4-1999, 4-2008, f-1999, f-2008, r-1999, r-2008)
# by 4 columns (cyl values). Strips on the right stack 2 lines: drv on top, year below.

Difficulty: Advanced

RYour turn

ex_5_5 <- # your code here ex_5_5

Click to reveal solution

RSolution

ex_5_5 <- ggplot(mpg, aes(displ, hwy)) + geom_point() + facet_grid(rows = vars(drv, year), cols = vars(cyl)) ex_5_5

Explanation: vars() accepts an arbitrary number of variables and stacks their strip labels in the panel margin. The formula syntax drv + year ~ cyl is equivalent but less explicit; vars() is the preferred current style. Each additional faceting variable multiplies the panel count, so check dplyr::n_distinct() before committing or you can produce hundreds of micro-panels that no longer fit on screen.

Navigate

Tidyverse packages

Deep dives

Wrangling & EDA

Statistics

Machine Learning

Time Series

By Industry

Reporting & Apps

Levels

ggplot2 Facets Exercises in R: 20 Practice Problems

Section 1. facet_wrap fundamentals (4 problems)

Exercise 1.1: Split a scatterplot into one panel per car class

Exercise 1.2: Force a two-row layout with nrow

Exercise 1.3: Let each panel pick its own y-axis range

Exercise 1.4: Rename facet strip labels without renaming the column

Section 2. facet_grid layouts (4 problems)

Exercise 2.1: Cross-classify points with facet_grid

Exercise 2.2: Stack panels in a single column with the dot placeholder

Exercise 2.3: Show marginal totals with facet_grid margins

Exercise 2.4: Free the y-axis per row in a facet_grid

Section 3. Scales, space, and panel order (3 problems)

Exercise 3.1: Allocate panel height proportional to bar count

Exercise 3.2: Fix the lower y-bound but free the upper bound per panel

Exercise 3.3: Reorder facet panels by a summary statistic

Section 4. Strip labels and styling (4 problems)

Exercise 4.1: Style facet strips with a custom theme

Exercise 4.2: Wrap long facet labels onto multiple lines

Exercise 4.3: Show row and column variable names together with label_both

Exercise 4.4: Move facet strips to the bottom of each panel

Section 5. Faceting in real workflows (5 problems)

Exercise 5.1: Small multiples for a long-format time series

Exercise 5.2: Facet a continuous variable by binning

Exercise 5.3: Highlight one panel by greying the rest

Exercise 5.4: Add a per-panel summary statistic as a label

Exercise 5.5: Use facet_grid with two row variables via vars()

What to do next

ggplot2 Facets Mastery