kaizenstat 0.2.0

Zero-friction AutoML + Data Cleaning Toolkit

🚀 KaizenStat

KaizenStat is a zero-friction data validation, automatic cleaning, and AutoML benchmarking toolkit. Diagnose datasets instantly, auto-repair issues, train baseline models, generate standalone Python code, and launch interactive dashboards — all in one command.

---

✨ Features

Command	What it does
kz audit	🔍 Diagnostic sweep — duplicates, NaNs, infs, ID columns, imbalance
kz heal	🩹 Auto-clean — impute, deduplicate, drop dead columns
kz benchmark	🚀 Train & rank ML models with cross-validation
kz auto	⚡ Full pipeline in one command (audit → heal → benchmark)
kz explain	💬 Plain-English summary of findings and recommendations
kz codegen	📝 Generate a standalone Python training script
kz export-model	💾 Train best model and save to .joblib
kz report	📊 Generate interactive HTML report with charts
kz serve	🌐 Launch interactive Streamlit web dashboard

---

📦 Installation

pip install kaizenstat

Optional extras:

pip install kaizenstat[ui]     # + Streamlit dashboard
pip install kaizenstat[gpu]    # + XGBoost GPU support
pip install kaizenstat[fast]   # + Polars fast data loading
pip install kaizenstat[all]    # everything

---

🚀 Quick Start

Python API

from kaizenstat import KaizenStat

# Full pipeline in one call
KaizenStat.auto("data.csv", target="price")

# Or step-by-step
import pandas as pd
df = pd.read_csv("data.csv")

KaizenStat.audit(df, target="price")
df_clean = KaizenStat.heal(df, target="price")
results = KaizenStat.benchmark(df_clean, target="price")

💬 Get a Plain-English Explanation

KaizenStat.explain("data.csv", target="price")

📝 Generate Standalone Code

KaizenStat.codegen("data.csv", target="price", output_path="deploy.py")

💾 Export & Load Models

# Train + save
KaizenStat.auto("data.csv", target="price")
KaizenStat.save_model(path="model.joblib")

# Load later
pipeline = KaizenStat.load_model("model.joblib")
predictions = pipeline.predict(new_data)

📊 Generate HTML Report

KaizenStat.report("data.csv", target="price", output_path="report.html")

🌐 Launch Web Dashboard

KaizenStat.serve("data.csv", target="price")

---

💻 CLI Usage

# Diagnostic sweep
kz audit data.csv --target price

# Auto-clean dataset
kz heal data.csv --target price -o clean.csv

# Train & rank models
kz benchmark clean.csv --target price

# Full pipeline
kz auto data.csv --target price

# Plain-English summary
kz explain data.csv --target price

# Generate standalone Python script
kz codegen data.csv --target price -o deploy.py

# Train best model and export
kz export-model data.csv --target price -o model.joblib

# Generate interactive HTML report
kz report data.csv --target price -o report.html

# Launch web dashboard
kz serve data.csv --target price

---

🛠 Development

git clone https://github.com/yourusername/kaizenstat.git
cd kaizenstat
pip install -e ".[all]"

---

📄 License

Distributed under the MIT License.