facets-overview

What it is and what it does

Facets Overview is a Python library that computes and visualizes summary statistics for dataset features in an interactive Jupyter notebook interface. It accepts data as pandas DataFrames or TensorFlow Example protocol buffers from TfRecord files, then generates a protocol buffer containing statistics like min, mean, median, max, and standard deviation for numeric features, and metrics like average length, unique value counts, and mode for string features. The visualization displays these statistics in two tables—one for numeric and one for categorical features—with sortable rows, distribution charts, and optional weighted statistics if example weights are provided.

The package is designed for exploratory data analysis and data quality assessment in machine learning workflows. It highlights potentially problematic statistics (such as missing values) in red and offers multiple chart types including histograms, deciles, and cumulative distribution functions. However, the package is abandoned and has not been updated since May 2023, meaning it may face compatibility issues with newer versions of its dependencies (numpy, pandas, protobuf) or modern Jupyter environments.

Use it for:

• Explore feature distributions and detect data quality issues in a pandas DataFrame before training an ML model.
• Compare summary statistics across multiple datasets side-by-side to identify distribution shifts or anomalies.
• Generate weighted statistics for datasets where examples have importance weights, toggling between weighted and unweighted views.
• Analyze TensorFlow Example records from TfRecord files to understand feature statistics at scale in a TensorFlow pipeline.
• Create interactive reports of dataset characteristics in Jupyter notebooks for data documentation and team review.

Worth the install?

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

Generates summary statistics for dataset features and creates interactive visualizations of data distributions, supporting both numeric and categorical columns from pandas DataFrames or TensorFlow records.

Yes, but with caution. The package is useful for interactive data exploration and visualization in Jupyter notebooks, has low install friction, and carries no known vulnerabilities. However, it is abandoned and unmaintained since May 2023, so compatibility with current versions of numpy, pandas, protobuf, and Jupyter is uncertain. Install only if you can tolerate potential breakage or are working in a stable, locked environment. For active projects, consider maintained alternatives.

Install

facets-overview on PyPI

pip install facets-overview

uv add facets-overview

poetry add facets-overview

Installing facets-overview

Before you install

Low friction installation with a pure Python wheel, though the package is abandoned as of 2023-05-24 and has not been maintained for over 1178 days. It depends on numpy, pandas, and protobuf (version 3.20.0 or later required as of version 1.1.0), all widely available.

License in practice

Licensed under Apache 2.0, a permissive license that allows commercial and private use with minimal restrictions, making it safe to incorporate into most projects.

Quickstart

from facets_overview.generic_feature_statistics_generator import GenericFeatureStatisticsGenerator
import pandas as pd
df = pd.DataFrame({'num': [1, 2, 3, 4], 'str': ['a', 'a', 'b', None]})
proto = GenericFeatureStatisticsGenerator().ProtoFromDataFrames([{'name': 'test', 'table': df}])

Requires protobuf version 3.20.0 or later; TensorFlow is optional but needed only if analyzing TfRecord files rather than pandas DataFrames.

Verify before relying

• Whether the package works with modern versions of numpy, pandas, and protobuf despite being unmaintained since May 2023.
• Compatibility with current Jupyter notebook environments and whether the nbextension still installs correctly.
• Whether TensorFlow dependency handling remains functional given the rapid evolution of that ecosystem.

Package facts

Evidence: facets_overview-1.1.1-py2.py3-none-any.whl

Tags

More Information Analysis packages