How to Create a Heatmap in Python: matplotlib, seaborn, and plotly

Heatmaps turn a matrix of numbers into a colour-coded grid, making it easy to spot patterns, outliers, and clusters that would be invisible in a table. They appear in gene expression studies, correlation analysis, confusion matrices, and any dataset where two categorical axes cross a continuous measurement.

1. Basic heatmap with seaborn

sns.heatmap is the simplest starting point. Pass a 2-D array or DataFrame, set annot=True to show values, and choose a colourmap.

import seaborn as sns
import matplotlib.pyplot as plt
import numpy as np

rng = np.random.default_rng(42)
data = rng.uniform(0, 1, (8, 8))

fig, ax = plt.subplots(figsize=(7, 6))
sns.heatmap(data, annot=True, fmt=".2f", cmap="viridis",
            linewidths=0.4, linecolor="black", ax=ax)
ax.set_title("Heatmap with Annotations")
plt.tight_layout()

2. Correlation matrix (masked upper triangle)

For correlation matrices, mask the redundant upper triangle with np.triu. Set vmin=-1, vmax=1 and use the coolwarm diverging palette so the neutral zero reads as white.

import seaborn as sns
import matplotlib.pyplot as plt
import pandas as pd
import numpy as np

rng = np.random.default_rng(42)
df = pd.DataFrame(rng.standard_normal((100, 5)),
                  columns=["A", "B", "C", "D", "E"])
corr = df.corr()

mask = np.triu(np.ones_like(corr, dtype=bool))   # upper triangle mask

fig, ax = plt.subplots(figsize=(6, 5))
sns.heatmap(corr, mask=mask, annot=True, fmt=".2f",
            cmap="coolwarm", vmin=-1, vmax=1,
            square=True, linewidths=0.5, ax=ax)
ax.set_title("Correlation Matrix")
plt.tight_layout()

3. Clustered heatmap

sns.clustermap performs hierarchical clustering and reorders the rows and columns automatically. It creates its own figure, so do not pass ax manually.

import seaborn as sns
import matplotlib.pyplot as plt
import numpy as np

rng = np.random.default_rng(42)
data = rng.uniform(0, 1, (12, 8))

# clustermap handles its own figure — do not use fig, ax
g = sns.clustermap(data, cmap="magma", figsize=(8, 7),
                   row_cluster=True, col_cluster=True,
                   linewidths=0.2)
g.ax_heatmap.set_title("Clustered Heatmap")
plt.tight_layout()

Colourmap reference

Common mistakes

• Using rainbow colourmaps — jet and rainbow are not perceptually uniform and mislead readers. Use viridis or magma.
• No colourbar — always include a colourbar with a label showing the variable name and units.
• Annotating large matrices — annotation becomes unreadable above ~15×15 cells. Remove annot and rely on the colourbar instead.
• Wrong vmin/vmax on diverging maps — set both symmetrically around zero so the neutral colour maps correctly.