yaEDA: Yet Another EDA 🚀

yaEDA (Yet Another EDA) is an automated exploratory data analysis, feature intelligence, and model error diagnostics engine built for competitive tabular data science and machine learning.

Features

• Golden Feature Ranking: Combines tree MDI, out-of-sample permutation drop, and Mutual Information into a composite rank score.
• Partial Dependence & ICE Curves: Evaluates individual non-linear response curves across primary predictors.
• Arithmetic Synergy Engine: Evaluates combinations ($A \times B$, $A / B$, $A \pm B$) to identify feature interactions exceeding baseline performance.
• KMeans & PCA Clustering: Identifies distinct clusters and measures target alignment using mutual information.
• Model Error Diagnostics: Partitions false positive/negative cohorts and residual tails with global beeswarms and local SHAP waterfall plots.
• Self-Contained Dashboard: Exports standalone HTML reports with zero runtime dependencies.

Installation

# Core package
$ pip install yaeda

# With SHAP support
$ pip install "yaeda[shap]"

# With PDF export (WeasyPrint)
$ pip install "yaeda[pdf]"

Using uv:

$ uv add yaeda --extra shap

Quickstart

import pandas as pd
from yaeda import TabularEDA

df = pd.read_csv("train.csv")

eda = TabularEDA(
    df=df,
    target="target_col",
    target_type="classification",
)

# Generate offline HTML dashboard
eda.to_html("report.html")

# Export structured intelligence as JSON
eda.to_json("metadata.json")

Development Setup

# Clone the repository
git clone [https://github.com/your-username/yaEDA.git](https://github.com/your-username/yaEDA.git)
cd yaEDA

# Create virtual environment and sync dependencies using uv
uv sync --extra dev --extra all

# Run test suite
uv run pytest

# Run linting
uv run ruff check .