sweetviz

What it is and what it does

Sweetviz is a pandas-based Python library that automates exploratory data analysis by generating self-contained, interactive HTML reports from dataframes. It detects feature types automatically (numerical, categorical, text), computes summary statistics and associations, and visualizes relationships between features and optional target variables. The library is designed around two main workflows: analyzing a single dataset with an optional target feature, and comparing two datasets (e.g., training vs. test splits) or subsets within a dataset (e.g., male vs. female rows).

The package handles mixed-type associations seamlessly—Pearson correlation for numerical features, uncertainty coefficient for categorical features, and correlation ratio for categorical-numerical pairs. Output is a fully self-contained HTML application that renders in a browser or notebook environment, making it useful for quick data characterization tasks without manual visualization code. It depends on standard scientific Python libraries (pandas, numpy, scipy, matplotlib) and template rendering (jinja2).

Use it for:

• Generate a quick summary report of a new dataset to understand distributions, missing values, and feature types without writing custom code.
• Compare training and test datasets side-by-side to identify data drift or distribution mismatches before model training.
• Analyze how a target variable (e.g., loan default, customer churn) relates to other features to guide feature engineering.
• Visualize correlations and associations across mixed data types to identify potential multicollinearity or feature interactions.
• Create an interactive HTML report for stakeholders to explore data characteristics without requiring Python knowledge.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Sweetviz generates interactive HTML visualizations for exploratory data analysis (EDA) in two lines of code, supporting target analysis, dataset comparison, and correlation detection across numerical, categorical, and mixed-type features.

Yes. Sweetviz is actively maintained, permissively licensed, has low install friction, and solves a common pain point in data science workflows. It is well-suited for rapid EDA when you need quick visual summaries of datasets and their relationships. Install it if you work regularly with pandas dataframes and want to reduce boilerplate visualization code; skip it if you need fine-grained control over plot aesthetics or work primarily in restricted environments.

Install

sweetviz on PyPI

pip install sweetviz

uv add sweetviz

poetry add sweetviz

Installing sweetviz

Before you install

Low install friction with a wheel distribution and eight common dependencies (pandas, numpy, matplotlib, scipy, jinja2, tqdm, and importlib utilities). Actively maintained as of April 2026 with recent bug fixes in version 2.3.2.

License in practice

MIT license permits free use, modification, and distribution with minimal restrictions, making it suitable for both open-source and commercial projects.

Quickstart

import sweetviz as sv
import pandas as pd

df = pd.read_csv('data.csv')
my_report = sv.analyze(df)
my_report.show_html()

Requires pandas 0.25.3 or later and Python 3.7+; custom file operations in restricted environments like Google Colab may not be fully supported.

Verify before relying

• Whether version 2.3.3 resolves all long-standing issues mentioned in the April 2026 update beyond what 2.3.2 addressed.
• Current status of Google Colab and other restricted-environment support mentioned as 'looking into a solution'.
• Performance characteristics for datasets with very large feature counts relative to the association_auto_threshold parameter.

Package facts

Evidence: sweetviz-2.3.3-py3-none-any.whl

Keywords: pandas, data-science, data-analysis, python, eda

Tags

More Python Modules packages