Heatmap

Seaborn basics

2 min read

Published Oct 7 2025

24
0
0
0
ChartsGraphsMatplotlibNumPyPandasPythonSeabornVisualisation

seaborn.heatmap() creates a colour-coded matrix (grid) to visualise two-dimensional data.
Each cell’s colour represents the value in that position — making it ideal for:

  • Correlation matrices
  • Pivot tables
  • Confusion matrices
  • Any 2D numerical array or DataFrame

Syntax:

sns.heatmap(
    data,
    *,
    vmin=None,
    vmax=None,
    cmap=None,
    center=None,
    robust=False,
    annot=False,
    fmt=".2g",
    annot_kws=None,
    linewidths=0,
    linecolor="white",
    cbar=True,
    cbar_kws=None,
    square=False,
    xticklabels=True,
    yticklabels=True,
    mask=None,
    ax=None,
    **kwargs
)

Parameters:

  • data = 2D dataset (DataFrame, array, or matrix)
  • vmin, vmax = Value range for colour scaling
  • cmap = Colour map (e.g., "viridis", "coolwarm", "magma")
  • center = Value where colourmap is centered (useful for diverging data)
  • annot = Annotate cells with numeric values
  • fmt = Format string for annotations (e.g., ".1f")
  • linewidths, linecolor = Control grid lines between cells
  • cbar = Show/hide colour bar
  • cbar_kws = Arguments for colour bar customisation
  • square = Make cells square-shaped
  • xticklabels, yticklabels = Show or hide tick labels
  • mask = Boolean mask to hide specific cells




Basic example

import seaborn as sns
import matplotlib.pyplot as plt

data = sns.load_dataset("flights").pivot(index="month", columns="year", values="passengers")

sns.heatmap(data)
plt.show()

Creates a heatmap of passengers per month per year. Brighter colours indicate higher values.


seaborn heatmap plot basic example





Add annotations

import seaborn as sns
import matplotlib.pyplot as plt

data = sns.load_dataset("flights").pivot(index="month", columns="year", values="passengers")

sns.heatmap(data, annot=True, fmt=".0f")
plt.show()

Displays each cell’s numeric value inside the heatmap. fmt controls the text formatting (here, integer).


seaborn heatmap plot annotations example





Change colour palette

import seaborn as sns
import matplotlib.pyplot as plt

data = sns.load_dataset("flights").pivot(index="month", columns="year", values="passengers")

sns.heatmap(data, cmap="YlGnBu") # Yellow-Green-Blue
plt.show()

Colour map options include:

  • "viridis" (default)
  • "coolwarm"
  • "magma"
  • "cividis"
  • "crest", "flare", "icefire", etc.

Try sns.color_palette() to preview available palettes.


seaborn heatmap plot colours example





Add lines between cells

import seaborn as sns
import matplotlib.pyplot as plt

data = sns.load_dataset("flights").pivot(index="month", columns="year", values="passengers")

sns.heatmap(data, linewidths=1, linecolor="white")
plt.show()

Adds visible grid lines between cells for clarity.


seaborn heatmap plot gridlines example





Centre the colourmap

import seaborn as sns
import matplotlib.pyplot as plt

data = sns.load_dataset("flights").pivot(index="month", columns="year", values="passengers")

sns.heatmap(data, cmap="coolwarm", center=data.mean().mean())
plt.show()

Centres colours around a specific value (e.g., the dataset mean). Useful for highlighting deviations above or below a midpoint.


seaborn heatmap plot centre example





Customise the colour scale range

sns.heatmap(data, vmin=100, vmax=600)

Fixes the colour scale between specific min and max values. Ensures consistent comparison across multiple heatmaps.






Hide the colour bar

import seaborn as sns
import matplotlib.pyplot as plt

data = sns.load_dataset("flights").pivot(index="month", columns="year", values="passengers")

sns.heatmap(data, cbar=False)
plt.show()

Removes the colour scale legend on the side.


seaborn heatmap plot no colourbar example





Square cells

import seaborn as sns
import matplotlib.pyplot as plt

data = sns.load_dataset("flights").pivot(index="month", columns="year", values="passengers")

sns.heatmap(data, square=True, cmap="YlOrBr")
plt.show()

Forces the cells to be square-shaped (1:1 aspect ratio). Helpful when displaying symmetric matrices (e.g., correlations).


seaborn heatmap plot square example





Show only part of the matrix

import seaborn as sns
import matplotlib.pyplot as plt

data = sns.load_dataset("flights").pivot(index="month", columns="year", values="passengers")

# mask cells with value < 300
mask = data < 300

sns.heatmap(data, mask=mask, cmap="coolwarm")
plt.show()

Masks (hides) cells based on a condition. Useful for focusing on higher or lower values only.


seaborn heatmap plot mask example





Hide axis labels

import seaborn as sns
import matplotlib.pyplot as plt

data = sns.load_dataset("flights").pivot(index="month", columns="year", values="passengers")

sns.heatmap(data, xticklabels=False, yticklabels=False)
plt.show()

Removes axis tick labels for a cleaner look.


seaborn heatmap plot hide axis labels example





Correlation matrix example

import seaborn as sns
import matplotlib.pyplot as plt

iris = sns.load_dataset("iris")

corr = iris.corr(numeric_only=True)
sns.heatmap(corr, annot=True, cmap="coolwarm", square=True)
plt.title("Correlation Matrix (Iris Dataset)")
plt.show()

Each cell shows the correlation between two numeric variables. Useful for identifying strong relationships or multicollinearity.


seaborn heatmap plot correlation example





Confusion matrix example

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

conf_matrix = pd.DataFrame(
    [[50, 2, 1],
     [3, 45, 5],
     [0, 4, 48]],
    index=["Class A", "Class B", "Class C"],
    columns=["Pred A", "Pred B", "Pred C"]
)

sns.heatmap(conf_matrix, annot=True, fmt="d", cmap="Blues")
plt.title("Confusion Matrix")
plt.show()

Shows true vs. predicted labels for classification models. The diagonal cells represent correct predictions.


seaborn heatmap plot confusion example





Diverging palette example

import seaborn as sns
import matplotlib.pyplot as plt

iris = sns.load_dataset("iris")

corr = iris.corr(numeric_only=True)
sns.heatmap(corr, annot=True, cmap="RdBu_r", center=0)
plt.title("Centered Correlation Heatmap")
plt.show()

Colours:

  • Blue = negative correlations
  • Red = positive correlations
  • White = near zero (neutral relationship)

seaborn heatmap plot diverging example





Horizontal colour bar

import seaborn as sns
import matplotlib.pyplot as plt

data = sns.load_dataset("flights").pivot(index="month", columns="year", values="passengers")

sns.heatmap(
    data,
    cmap="YlGnBu",
    cbar_kws={"orientation": "horizontal", "shrink": 0.8, "label": "Passenger Count"}
)
plt.show()

Moves the color bar below the heatmap and adds a label.


seaborn heatmap plot horizontal colourbar example
© 2025 SimpleSteps.guide
AboutFAQPoliciesContact