percentify

What it is and what it does

Percentify is a data diagnostics library that wraps common exploratory and quality-check operations into single-call functions for pandas and Polars DataFrames. Its flagship function, `profiler()`, scans a DataFrame for data issues—missing values, outliers, collinearity, class imbalance, skew—ranks them by severity, and suggests fixes. The library also provides functions for variance analysis (cv, pca_variance, pca_loadings), statistical testing (permutation_test, bootstrap_ci, effect_size), correlation detection, and formatting utilities.

The package is designed around a specific philosophy: each function returns the single most common answer in one call, with clear output sorted worst-first, and points users to underlying libraries (pandas, scipy, statsmodels, scikit-learn) when deeper customization is needed. It treats pandas and Polars as first-class backends—pass either type and get the same type back without flags or manual conversion. Runtime dependencies are numpy, pandas, and scipy.

Use it for:

• Run `profiler()` before modeling to catch data issues and get a 0-100 health score for a CI data-quality gate.
• Use `missing()` to quickly see which columns have gaps and how much data is lost per column.
• Call `vif()` to detect multicollinearity and identify which features to drop before regression.
• Apply `outliers()` to find the percentage of outliers in each column and decide on treatment.
• Use `correlate()` to find feature pairs that move together and test whether the relationship is statistically significant.
• Call `imbalance()` on a target column to measure class skew and decide on resampling or weighting strategies.

Worth the install?

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

Percentify provides one-call exploratory statistics and data-quality diagnostics for pandas and Polars DataFrames, ranking issues worst-first with suggested fixes.

Yes. Percentify fills a real gap: it automates the 80% of exploratory checks you run on every dataset into one or two function calls, with low install friction, active maintenance, permissive licensing, and genuine dual support for pandas and Polars. No known vulnerabilities. Best for teams doing frequent data intake and quality validation.

Install

percentify on PyPI

pip install percentify

uv add percentify

poetry add percentify

Installing percentify

Before you install

Low install friction with a pure-Python wheel and three well-established runtime dependencies (numpy, pandas, scipy). Active maintenance with a release 32 days ago.

License in practice

Apache-2.0 is permissive; you can use this in commercial and proprietary projects with minimal restrictions, provided you include a copy of the license.

Quickstart

pip install percentify

import pandas as pd
from percentify import profiler

df = pd.DataFrame({"salary": [50000, None, 60000], "age": [25, 30, None]})
report = profiler(df)
print(report.to_frame())  # ranked findings with fixes
print(report.health)      # 0-100 data-health score

Requires Python 3.10 or later and pandas 2.0+.

Verify before relying

• Whether the package's Polars support is truly feature-parity with pandas across all functions.
• Performance characteristics on large datasets (memory usage, runtime scaling).
• Availability and completeness of the documentation at data-centt.github.io/percentify.

Package facts

Evidence: percentify-1.0.2-py3-none-any.whl

Keywords: data-science, statistics, pandas, eda, multicollinearity, vif, pca, outliers, data-analysis

Tags

More Information Analysis packages