Your First Chart

We're going to slow down and build a single chart together, one decision at a time. You've seen many charts so far; this page is about the process of arriving at one. By the end, you should feel ready to make a chart from any dataset placed in front of you.

The data

We'll use Plotly's built-in tips dataset — restaurant tip data collected by a single waiter over a few weeks. It is small, relatable, and has both numeric and categorical columns.

You should see columns like total_bill, tip, sex, smoker, day, time, size.

The question

Before any chart, ask: what question am I trying to answer? Let's pick a concrete one:

Do bigger restaurant bills lead to bigger tips, and does it matter whether the customer is a smoker?

This is two questions, really:

1. Is there a relationship between total_bill and tip?
2. Does that relationship differ for smokers vs non-smokers?

Choosing the chart type

Walk through the data types:

• total_bill — quantitative.
• tip — quantitative.
• smoker — categorical (Yes / No).

When we have two quantitative variables, a scatter plot is almost always the right starting point. (Lines connect points; we only want lines when there's a meaningful order, like time.)

The third variable (smoker) is categorical and small (just two groups), so we'll map it to color.

Building it up step by step

We'll add encodings one at a time so you can see each one earn its place.

Step 1: bare scatter

This is the absolute minimum. We can already see that yes, bigger bills tend to come with bigger tips — there's a clear upward cloud. But the chart is anonymous: no title, no axis labels, no groups.

Step 2: add color for smokers

Now we can compare. Hover and toggle the legend (click "Yes" or "No"). Do the two clouds look different? Maybe slightly; nothing dramatic.

Step 3: add a regression line per group

With trendline="ols", Plotly Express fits an ordinary-least- squares line to each color group. Suddenly the question "is the relationship different?" gets a visual answer: the two lines' slopes are similar — bigger bills predict bigger tips at roughly the same rate for both groups.

Step 4: title, labels, and the simple_white template

A chart without labels is rude to its readers.

The chart now communicates without needing any external context:

• The title says what's being shown.
• The axis labels carry units (USD).
• The legend has a friendly label ("Smoker?").
• The template (simple_white) gives a clean look that won't fight the data.

Step 5: a finishing touch — hover information

The default hover already shows the x and y values. Let's enrich it with the party size and the day, so a single hover tells the whole story of a row.

Now hover over any dot. You'll see the bill, tip, smoker status, day, time, and party size in one tooltip. The chart has become a small interactive analytical tool, not just a picture.

The pattern, summarized

Every chart you build in this course follows roughly the same recipe:

The loop closes: a good chart almost always raises a new question, which sends you back to the top of the loop.

Common first-chart mistakes

1. Skipping the question. "I'll just plot it and see" can be exploratory, but never call that chart "the answer." Always know what you're asking.
2. No labels. A chart with axes x and y is unfinished.
3. Default styling. The default Plotly template has a gray background and gridlines that fight your data. Use template="simple_white" from the start.
4. Encoding everything. Color, size, shape, faceting — all at once produces visual mush. Add encodings one at a time and stop when the question is answered.

Try it yourself

Modify the chart below to:

1. Use x="size" (party size) and y="tip" instead.
2. Color by day instead of smoker.
3. Update the title and labels accordingly.

Check your understanding

QuestionSelect one

What should be the first step when starting a new chart?

Pick a color palette.

Pick a chart type.

Decide what question the chart will answer.

Pick a template.

QuestionSelect one

You have two quantitative variables and want to see if they're related. Which chart type is the natural starting point?

Pie chart.

Bar chart.

Scatter plot (px.scatter).

Heatmap.

QuestionSelect one

Why is labels={...} worth adding to a Plotly Express call?

It makes the chart load faster.

It changes the data values.

It overrides the raw column names with human-readable labels (e.g., "Total bill (USD)" instead of "total_bill") for axes, legend, and hover.

It hides the legend.