Charts in Python: Matplotlib vs Plotly

Matplotlib and Plotly are the two most popular ways to make charts in Python. Use Matplotlib when you need static, precisely controlled images for reports, slides and PDFs; use Plotly when you want interactive charts with hover details, zoom and filtering for the browser, notebooks or web dashboards. Most analysts end up using both, and this guide shows the same charts in each so you can decide quickly.

Matplotlib vs Plotly at a glance

Both are free and open source. Install them with pip install matplotlib plotly; Anaconda includes them already. Many teams standardize on one library for reports and the other for exploration, which is a perfectly reasonable split. You do not need to master every option in either library; a handful of chart types (line, bar, scatter, histogram and box plot) covers most business questions.

Sample data for both libraries

All examples use a small pandas DataFrame of monthly sales by region. Replace it with pd.read_csv("your_file.csv") when you are ready.

import pandas as pd
import numpy as np

rng = np.random.default_rng(42)
df = pd.DataFrame({
    "Month": pd.date_range("2025-01-01", periods=12, freq="MS").repeat(2),
    "Region": ["East", "West"] * 12,
    "Revenue": rng.integers(20_000, 60_000, size=24),
})
monthly = df.groupby("Month", as_index=False)["Revenue"].sum()

If pandas is new to you, our pandas tutorial for Excel users explains groupby and pivot_table.

Matplotlib: static charts with full control

Line chart

import matplotlib.pyplot as plt
from matplotlib.ticker import StrMethodFormatter

fig, ax = plt.subplots(figsize=(9, 4.5))
ax.plot(monthly["Month"], monthly["Revenue"], marker="o", linewidth=2)
ax.set_title("Monthly revenue")
ax.set_ylabel("Revenue")
ax.yaxis.set_major_formatter(StrMethodFormatter("${x:,.0f}"))
ax.spines[["top", "right"]].set_visible(False)
fig.tight_layout()
fig.savefig("monthly_revenue.png", dpi=200)
plt.show()

Notice the object-oriented style: you create a figure (fig) and axes (ax) and call methods on them. It is slightly longer than the quick plt.plot() style, but it scales cleanly to multi-chart layouts and is what most experienced users recommend.

Stacked bar chart by region

by_region = df.pivot_table(index="Month", columns="Region",
                           values="Revenue", aggfunc="sum")
by_region.index = by_region.index.strftime("%b")

fig, ax = plt.subplots(figsize=(9, 4.5))
by_region.plot(kind="bar", stacked=True, ax=ax, rot=0)
ax.set_title("Revenue by region")
ax.set_xlabel("")
fig.tight_layout()
plt.show()

Converting the date index to short month labels before plotting keeps the bar chart's axis tidy. The pandas .plot() method uses Matplotlib under the hood, so you can always fine-tune the result through ax.

Small multiples: one chart per region

fig, axes = plt.subplots(1, 2, figsize=(11, 4), sharey=True)
for ax, region in zip(axes, ["East", "West"]):
    data = df[df["Region"] == region]
    ax.plot(data["Month"], data["Revenue"], marker="o")
    ax.set_title(region)
    ax.tick_params(axis="x", rotation=45)
fig.suptitle("Monthly revenue by region")
fig.tight_layout()
plt.show()

Small multiples are often easier to read than one crowded chart with many lines. sharey=True puts both panels on the same scale, so readers can compare regions honestly instead of being misled by different axis ranges.

Plotly: interactive charts in a few lines

Line chart

import plotly.express as px

fig = px.line(monthly, x="Month", y="Revenue", markers=True,
              title="Monthly revenue")
fig.update_layout(yaxis_tickprefix="$", yaxis_tickformat=",.0f")
fig.show()
fig.write_html("monthly_revenue.html")

Stacked bar chart by region

fig = px.bar(df, x="Month", y="Revenue", color="Region",
             title="Revenue by region")
fig.show()

Plotly Express takes the "long" DataFrame directly and stacks the bars by color automatically, so no pivot is needed. Hover over any bar to see its exact value, click a legend item to hide a region, and drag to zoom. The saved HTML file opens in any browser, which makes it easy to share an interactive chart with someone who does not use Python.

Exporting a Plotly chart as a static PNG requires the extra kaleido package (pip install kaleido) and then fig.write_image("chart.png").

Customizing hover labels and reference lines

avg = df["Revenue"].mean()  # average per region per month

fig = px.line(df, x="Month", y="Revenue", color="Region",
              markers=True, template="plotly_white")
fig.add_hline(y=avg, line_dash="dash", annotation_text="Average")
fig.update_traces(hovertemplate="%{x|%b %Y}<br>$%{y:,.0f}")
fig.show()

hovertemplate controls exactly what appears in the tooltip, and add_hline draws a reference line such as a target or average. The average is calculated from the same per-region monthly values the lines show, so the comparison is fair.

How to choose: a simple decision guide

1. Where will the chart live? A PDF, printed report or Word document points to Matplotlib. A browser, notebook or web app points to Plotly.
2. Will people explore it? If readers need to hover for exact values or zoom into a date range, choose Plotly.
3. Is it going into Power BI? Power BI's Python visual renders static images through Matplotlib (or libraries built on it, such as seaborn). See Python and R Visuals in Power BI.
4. Are you building a dashboard app? Plotly integrates naturally with Dash and Streamlit; see Build Web Dashboards with Streamlit.
5. Do you need a publication-quality figure with exact styling? Matplotlib gives you control over every pixel.

Whichever library you pick, the design principles are the same: one clear message per chart, a descriptive title, labeled axes, minimal clutter and color used with purpose.

Common chart mistakes in both libraries

• Truncated bar axes: bar charts should start at zero; otherwise small differences look dramatic. Matplotlib and Plotly both usually start bars at zero, but check after you set custom limits.
• Too many colors: more than five or six series in one chart becomes hard to read. Group minor categories into "Other" or switch to small multiples.
• Unformatted numbers: raw values like 1234567.891 on an axis slow readers down. Use the tick formatting shown above.
• Default titles: a title such as "Revenue" says little. A title like "West revenue fell in Q3" tells the reader what to notice.

For more on these principles, see Data Visualization Principles That Actually Work.

Where seaborn fits

seaborn is a statistical plotting library built on Matplotlib. It produces attractive histograms, box plots, heatmaps and regression plots with very little code, for example sns.boxplot(data=df, x="Region", y="Revenue"). Because its output is a Matplotlib figure, you can customize it with the same ax methods shown above and use it anywhere Matplotlib works, including Power BI Python visuals.

Next steps

Take one chart you rebuild by hand every month and recreate it in both libraries using the code above. Save the Matplotlib version as a PNG for your report and the Plotly version as HTML for colleagues who like to explore. Once you know which one your audience prefers, turn the script into a reusable function so next month's chart takes seconds.

Watch: video lessons

▶

▶

Frequently asked questions