Who This Is For

This article is for readers who can already draw a few plots but still feel that ggplot2 behaves like magic. The goal here is to make the grammar visible: what a layer is, what mapping means, what setting means, and why different positions change the same data into different visuals.

What You Will Do

• Build one plot in stages.
• Compare mapping and setting.
• Compare stack, dodge, and jitter.
• Understand how a ggplot object grows layer by layer.

Before You Start

• You should already be comfortable with basic ggplot() and a few simple geom_*() calls.
• You need ggplot2, dplyr, and patchwork.
• You should be willing to read one plot as a combination of decisions rather than a single command.

The companion script for this article is:

R draw/scripts/04-ggplot-from-zero-how-ggplot-builds-a-plot.R

Step 1: Build a Plot in Stages

Here is the most important grammar idea in ggplot2:

ggplot(data, aes(...)) + layer_1 + layer_2 + layer_3

That means you are not issuing one magical plotting command. You are assembling a plot object piece by piece.

The companion script saves a side-by-side comparison of a simpler stage and a richer stage:

The left panel shows a first layer with points. The right panel adds a mapped aesthetic and a smoothing layer. This is how a ggplot grows.

Step 2: Understand Mapping vs Setting

One of the most important beginner distinctions is this:

• mapping happens inside aes()
• setting happens outside aes()

Example:

ggplot(mpg, aes(x = drv, fill = class)) +
  geom_bar()

Here fill = class maps a variable from the data.

Now compare it with this:

ggplot(mpg, aes(x = drv, fill = class)) +
  geom_bar(fill = "#2C7FB8")

Now fill is being set to one fixed color. That means the grouped fill mapping is no longer visually visible.

Step 3: Learn Position Adjustments

The same bar or point geometry can communicate different things depending on position.

The script compares three common choices:

Use them like this:

• stack: default for grouped bar counts
• dodge: place groups side by side
• jitter: reduce overlap in points

These are not cosmetic changes. They change how your reader compares the data.

Step 4: Use the Grammar as a Thinking Tool

When a plot looks wrong, ask:

1. Did I choose the right data?
2. Did I map the right variables in aes()?
3. Did I pick the right geometry?
4. Did I accidentally set a constant where I meant to map a variable?
5. Is the position adjustment helping or hiding the comparison?

That checklist is often more useful than memorizing more syntax.

How to Confirm It Worked

• Your script creates:
  • R draw/figures/04-build-stages.png
  • R draw/figures/04-position-adjustments.png
  • R draw/figures/04-wrong-vs-correct.png
• You can explain the difference between mapping and setting without looking at notes.
• You can explain why the same bar chart data looks different under stack and dodge.

Common Questions

Why does ggplot use + between layers?

Because each new layer is added to an existing plot object. That design reflects the grammar: plots are assembled from components.

Is geom_smooth() a new plot type?

No. It is a new layer added to an existing plot. The base scatter plot is still there underneath.

Why not teach every geom before teaching grammar?

Because without the grammar, beginners often memorize disconnected code. Grammar gives structure to everything that comes later.

Review Score

Score: 94/100 Verdict: This draft is ready for human review and gives the series its first strong conceptual backbone.

```

Personnel

• ✍ Creator: Chenglin Cai
• 🤖 AI Collaboration: ChatGPT
• 🧪 Data Provider: ggplot2 package dataset mpg
• 💻 Code Contributor: ChatGPT