🦈 SharkPy

A friendly machine learning framework with shark-themed feedback! SharkPy simplifies the machine learning workflow while making it fun and educational.

Features

• Model Battle Arena: Compare multiple models automatically
• Smart Reporting: Generate reports in TXT, PDF, and DOCX formats
• Interactive Visualization: Beautiful plots with shark-themed styling
• Model Explanations: Clear explanations of model behavior
• Automated Optimization: Hyperparameter tuning with Optuna
• Shapash Integration: Interactive dashboards for model interpretation

Quick Start

Installation

# Basic installation
pip install sharkpy

# Full installation (includes LightGBM and CatBoost)
pip install sharkpy[full]

# Development installation
pip install sharkpy[dev]

Basic Usage

from sharkpy import Shark
import pandas as pd

# Create a Shark instance
shark = Shark()

# Load your data
data = pd.read_csv('your_data.csv')

# Train a model
shark.learn(
    data=data,
    target='target_column',
    model_choice='random_forest'
)

# Make predictions
predictions = shark.predict(new_data)

# Generate reports
shark.report(export_path='report.pdf', format='pdf')

Model Battle Example

# Define models to compete
models = [
    'linear_regression',
    'random_forest',
    'xgboost',
    'lightgbm',
    'catboost'
]

# Let them battle!
battle_results = shark.battle(
    data=data,
    target='target_column',
    models=models,
    metric='r2'
)

# Get the champion
print(f"Winner: {battle_results['champion']}")
print(f"Score: {battle_results['score']:.4f}")

Supported Models

• Linear Regression
• Logistic Regression
• Random Forest
• Support Vector Machines
• Ridge Regression
• Lasso Regression
• K-Nearest Neighbors
• XGBoost
• LightGBM (with full installation)
• CatBoost (with full installation)

Reports

SharkPy can generate comprehensive reports in multiple formats:

# Text report
shark.report(export_path='report.txt', format='txt')

# PDF report (requires MS Word or LibreOffice)
shark.report(export_path='report.pdf', format='pdf')

# Word document
shark.report(export_path='report.docx', format='docx')

Visualizations

# Prediction plot
shark.plot(kind="prediction")

# Feature importance
shark.plot(kind="feature_importance")

# Interactive Shapash dashboard
shark.explain_with_shapash()

Model Explanations

# Get friendly explanations of your model
shark.explain()

# Create interactive Shapash dashboard
shark.explain_with_shapash(title_story="My Analysis")

Save & Load Models

# Save your model
shark.save_model(name="my_model")

# Load a saved model
shark.load_model("models/my_model.joblib")

Development

# Install development dependencies
pip install sharkpy[dev]

# Run tests
pytest

# Format code
black sharkpy/

Documentation

Full documentation is available at sharkpy.readthedocs.io

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• Built with scikit-learn, XGBoost, and other amazing open-source tools
• Inspired by making machine learning more accessible and fun

Contact

• Author: Ezz Eldin Ahmed
• Email: ezzeldinahmad96@gmail.com
• GitHub: Ezzio11

Made with 🦈 by SharkPy Team