The Story of Plotly

You will spend a lot of time with Plotly in the rest of this course, so it's worth knowing where it came from, what it actually is, and why Plotly Express in particular is the right tool for beginners.

A company, a library, a wrapper

The name "Plotly" can mean three different things, and people use them interchangeably in confusing ways. Let's be precise:

• Plotly (the company) — a Montréal-based company founded in 2012 by Alex Johnson, Jack Parmer, Chris Parmer, and Matthew Sundquist. Their original product was a web app for making charts collaboratively in the browser.
• Plotly.js — an open-source JavaScript charting library released by the company in 2015, built on top of D3.js and WebGL. It is what actually renders every Plotly chart you see.
• plotly.py — the Python wrapper around Plotly.js, also open source. It lets you build interactive web charts from Python.
• Plotly Express — a much simpler, higher-level API on top of plotly.py, released in 2019 by Nicolas Kruchten.

For most of this course, when we say "Plotly," we mean Plotly Express. It is the gateway drug. Almost every chart you need can be made with one line of Plotly Express. When you outgrow it, you can always drop down to plotly.graph_objects for full control — but that day is months away.

Why the company started

In 2012, the team at Plotly noticed something simple. Researchers and analysts spent enormous amounts of time making the same charts over and over again — in MATLAB, in Excel, in R, in Python — and then emailing screenshots of them around. Every chart's interactivity was lost as soon as it was shared.

Their early product was a web tool: paste a CSV, get back a chart URL you could share with collaborators, where everyone could still hover and zoom. It was a good idea, but it didn't catch on. What did catch on was the chart engine underneath it. Researchers wanted to embed the engine into their own notebooks and apps.

So Plotly open-sourced the engine. Plotly.js in 2015 was the result. It is, to this day, one of the most-downloaded JavaScript charting libraries in the world.

The "old" Python API

The first Python wrapper (plotly.py, around 2014) was powerful but verbose. A simple bar chart looked like this:

import plotly.graph_objects as go

fig = go.Figure()
fig.add_trace(go.Bar(
    x=["A", "B", "C"],
    y=[10, 7, 14],
    name="Counts",
))
fig.update_layout(
    title="Counts by category",
    xaxis_title="Category",
    yaxis_title="Count",
    template="simple_white",
)
fig.show()

That's nine lines — for the most basic chart in the world. The verbosity scared off beginners and frustrated even experienced analysts.

Plotly Express: one line, one chart

In 2019, Nicolas Kruchten released Plotly Express as a high-level API on top of plotly.graph_objects. The pitch was radical and simple:

A single function call, with a DataFrame as input, should produce a publication-quality interactive chart.

The bar chart above becomes:

import plotly.express as px

px.bar(df, x="category", y="count",
       title="Counts by category",
       template="simple_white")

That's one line of meaningful code. Try it:

The chart that came out is fully interactive. You can hover, zoom, download. There are no sacrifices for the brevity.

What makes Plotly Express different

Three design choices set Plotly Express apart from other Python chart libraries:

1. It takes a DataFrame and column names, not arrays. You write x="category", not x=df["category"]. This makes the code read more like a sentence about your data.
2. Encodings are arguments. Want to add color? Pass color="...". Want to facet into small multiples? Pass facet_col="...". The grammar is uniform across every chart type.
3. It produces interactive charts by default. Every output is pannable, zoomable, hoverable. You don't have to opt in.

Plotly Express's place in the ecosystem

You will hear about other Python chart libraries. Here's a fair summary:

• Matplotlib — the foundational, static plotting library. Excellent for publication (papers, books) and total customization. Verbose; not interactive by default.
• Seaborn — a high-level wrapper around Matplotlib, focused on statistical charts. Beautiful defaults; not interactive.
• Altair — a declarative chart library based on Vega-Lite. Elegant grammar; produces interactive web charts.
• Bokeh — another interactive web chart library, popular in scientific computing.
• Plotly Express — the easiest path from DataFrame to interactive chart. The choice for this course.

There is no single "best" library. Plotly Express is the right beginner library because it gives you the modern interactive experience with the fewest lines of code. Once you're comfortable, you can branch out — and many real teams use two or three of these libraries side by side.

A peek at what's possible

Here is a more substantial chart to give a sense of Plotly Express's range:

That's one Plotly Express call. The chart maps four variables at once (GDP, life expectancy, continent, population), uses a log scale, labels axes properly, and produces a fully interactive web chart. That density of meaning per line of code is what makes Plotly Express the right beginner tool.

Check your understanding