R Visualization Project: Reproduce 5 Real-World Charts from Scratch — Solved Step-by-Step

The fastest way to level up your ggplot2 skills is to reproduce charts you admire. This project gives you 5 real-world chart styles — from The Economist to scientific journals — and walks you through rebuilding each one from scratch using only ggplot2 and built-in R datasets.

How Do You Recreate The Economist's Signature Scatter Plot?

The Economist's data team produces some of the most recognizable charts in journalism — clean scatter plots with a light blue-gray background, minimal gridlines, and bold titles positioned above the plot area. In this first project, you'll reproduce that signature look using the mpg dataset.

That one block transforms a default scatter plot into something you'd see in a magazine. The secret ingredients are the light blue-gray background (#d5e4eb), white horizontal gridlines that float above it, and a carefully chosen color palette.

Let's break down the three key theme elements that do the heavy lifting.

The background and gridlines create the Economist's signature look. Setting both plot.background and panel.background to the same blue-gray means the chart and its margins blend together seamlessly. Then panel.grid.major.y = element_line(color = "white") draws white horizontal lines that appear to float.

The axis treatment is deliberately asymmetric. A thin dark line runs along the bottom x-axis (axis.line.x), while the y-axis has no line at all — only the white gridlines guide the eye. This pulls attention toward the data points rather than the chart frame.

The text hierarchy uses size and weight to create a visual pecking order. The title is bold and large, the subtitle is smaller and gray, and the caption sits quietly at the bottom left.

How Do You Build a FiveThirtyEight-Style Bar Chart From Scratch?

FiveThirtyEight's charts are the opposite of flashy — a light gray background, no axis lines, no borders, bold headlines, and muted single-color fills. The design philosophy is "get out of the way and let the data speak." Let's recreate that with a bar chart showing vehicle class counts.

Two things make this instantly recognizable as FiveThirtyEight. First, everything non-essential is removed — no axis lines, no tick marks, no axis titles. The bars and subtle gray gridlines carry all the information. Second, the headline does the work of an axis title — "SUVs Dominate" tells the story, so a y-axis label saying "Count" would be redundant.

The color choice is deliberate too. FiveThirtyEight typically uses a single strong accent color (#008fd5 is their signature blue) rather than mapping colors to categories. When every bar is the same color, the reader compares heights — which is the whole point of a bar chart.

How Do You Design a Journal-Ready Scientific Figure?

Scientific journals demand a specific aesthetic: white backgrounds, minimal decoration, black or gray color schemes, and properly labeled axes with units. Reviewers will reject figures with rainbow palettes or missing axis labels. Let's build one using the iris dataset.

Notice what's different from the Economist and FiveThirtyEight styles. There's no background color, no gridlines, and no headline — journal figures rely on figure captions below the chart, not titles inside it. The axis labels include units in parentheses because a reviewer will flag "Petal length" without "(cm)".

The grayscale palette isn't just an aesthetic choice — many journals still print in black and white. Using different shapes (scale_shape_manual) alongside color ensures the three species stay distinguishable even without color.

theme_classic() gives you the perfect starting point — a white background with axis lines and no gridlines. From there, you only need to adjust font sizes, add bold axis labels, and position the legend inside the plot area to save space.

How Do You Reproduce a New York Times Annotated Line Chart?

The New York Times data visualization team pioneered a style where annotations replace legends entirely. Instead of a color key off to the side, labels sit directly on or next to the data lines. The result is a chart that reads like a paragraph — your eye follows the line and picks up context as it goes. Let's recreate this with unemployment data.

Three techniques make this chart feel like the NYT. First, geom_rect() draws shaded bands behind the line to mark recession periods — the pink rectangles give instant context without requiring the reader to look up dates. Second, annotate("text") places labels directly on the chart instead of relying on a separate legend. Third, the bold peak label ("Peak: 15.4 million") tells the reader the most important number without making them trace the line to the exact point.

How Do You Create a Modern Lollipop Chart for Data Journalism?

Lollipop charts are bar charts' sleeker cousin — a thin segment topped with a dot. They work especially well when you have many categories with similar values, because the dots are easier to compare than thick bar edges. Modern data journalism sites use them frequently for rankings. Let's build one with the top 15 most fuel-efficient cars in mtcars.

The lollipop chart solves a specific problem: when you have 10+ categories, thick bars create visual clutter. The thin segment + dot combination reduces ink and draws the eye to the endpoints — which is where the actual comparison happens.

The highlight strategy is key to data journalism. Instead of coloring all 15 cars differently (which would overwhelm the reader), we use just two colors: red for the leader and blue for everyone else. The subtitle reinforces this — "the Toyota Corolla leads the pack" — so the visual and textual stories align.

Practice Exercises

Exercise 1: Economist-Style Box Plot

Create an Economist-style box plot showing highway MPG (hwy) by vehicle class from the mpg dataset. Requirements:

• Use the blue-gray background (#d5e4eb) with white horizontal gridlines
• Sort classes by median hwy (highest first) using reorder()
• Add a bold title: "Highway Fuel Economy Varies Widely by Vehicle Class"
• Add a subtitle and source caption
• Use a single fill color (#01a2d9) for all boxes

Exercise 2: Side-by-Side Panel — Two Styles, One Dataset

Create a two-panel figure using mpg data. The left panel is a FiveThirtyEight-style bar chart of average highway MPG by class. The right panel is a journal-style scatter plot of displ vs hwy. Arrange them side by side using gridExtra::grid.arrange() or by saving each plot to a variable and printing them.

Exercise 3: Annotated Time Series with Two Metrics

Using the economics dataset, create an NYT-style chart that shows both unemployment count (unemploy) and personal savings rate (psavert) over the 2000-2015 period. Since these have different scales, normalize both to a 0-1 range so they can share the y-axis. Add recession shading and direct labels for each line (no legend).

Complete Example

Let's build one more chart from a blank canvas to a polished final product, applying everything you've learned. We'll create an Economist-style grouped bar chart showing average highway MPG by vehicle class, comparing 1999 vs 2008 model years.

This chart combines several techniques from the project: the Economist blue-gray background, white gridlines, bold title hierarchy, and no axis titles (the subtitle does that job). The grouped bars use position_dodge() for side-by-side comparison, and the two-tone blue palette keeps the color scheme coherent with the background.

The title does something important — it tells the story ("Improved Modestly") rather than just describing the chart ("Highway MPG by Class and Year"). The reader knows the conclusion before they even look at the bars, and the bars serve as evidence.

Summary

The common thread across all five styles: remove everything that doesn't help the reader understand the data faster. Every gridline, axis tick, border, and color choice either earns its place by adding clarity — or it gets cut.

References

1. Wickham, H. — ggplot2: Elegant Graphics for Data Analysis, 3rd Edition. Springer (2024). Link
2. ggplot2 documentation — theme() reference. Link
3. Scherer, C. — A ggplot2 Tutorial for Beautiful Plotting in R (2019). Link
4. The R Graph Gallery — ggplot2 examples. Link
5. Schwabish, J. — Better Data Visualizations. Columbia University Press (2021).
6. Wilke, C. — Fundamentals of Data Visualization. O'Reilly (2019). Link
7. R Core Team — An Introduction to R. Link

Continue Learning

1. Publication-Quality Figures in R — The complete checklist for making ggplot2 figures journal- and thesis-ready.
2. ggplot2 Customization Exercises — 10 theme and scale practice problems to sharpen your styling skills.
3. ggplot2 Facet Exercises — 8 facet_wrap() and facet_grid() problems for multi-panel layouts.