anfis-toolbox 0.1.2

ANFIS Toolbox is a comprehensive Python library for creating, training, and deploying Adaptive Neuro-Fuzzy Inference Systems (ANFIS). It provides an intuitive API that makes fuzzy neural networks accessible to both beginners and experts.

ANFIS Toolbox

ANFIS Toolbox is a comprehensive Python library for creating, training, and deploying Adaptive Neuro-Fuzzy Inference Systems (ANFIS). It provides an intuitive API that makes fuzzy neural networks accessible to both beginners and experts.

🚀 Overview

• Takagi–Sugeno–Kang (TSK) ANFIS with the classic four-layer architecture (Membership → Rules → Normalization → Consequent).
• Regressor and classifier facades with a familiar scikit-learn style (fit, predict, score).
• Trainers (Hybrid, SGD, Adam, RMSProp, PSO) decoupled from the model for easy experimentation.
• 10+ membership function families. The primary public interfaces are ANFISRegressor and ANFISClassifier.
• Thorough test coverage (100%+).

📦 Installation

Install from PyPI:

pip install anfis-toolbox

🧠 Quick start

Regression

import numpy as np
from anfis_toolbox import ANFISRegressor

X = np.random.uniform(-2, 2, (100, 2))
y = X[:, 0]**2 + X[:, 1]**2

model = ANFISRegressor()
model.fit(X, y)
metrics = model.evaluate(X, y)

Classification

import numpy as np
from anfis_toolbox import ANFISClassifier

X = np.r_[np.random.normal(-1, .3, (50, 2)), np.random.normal(1, .3, (50, 2))]
y = np.r_[np.zeros(50, int), np.ones(50, int)]

model = ANFISClassifier()
model.fit(X, y)
metrics = model.evaluate(X, y)

🧩 Membership functions at a glance

• Gaussian (GaussianMF) - Smooth bell curves
• Gaussian2 (Gaussian2MF) - Two-sided Gaussian with flat region
• Triangular (TriangularMF) - Simple triangular shapes
• Trapezoidal (TrapezoidalMF) - Plateau regions
• Bell-shaped (BellMF) - Generalized bell curves
• Sigmoidal (SigmoidalMF) - S-shaped transitions
• Diff-Sigmoidal (DiffSigmoidalMF) - Difference of two sigmoids
• Prod-Sigmoidal (ProdSigmoidalMF) - Product of two sigmoids
• S-shaped (SShapedMF) - Smooth S-curve transitions
• Linear S-shaped (LinSShapedMF) - Piecewise linear S-curve
• Z-shaped (ZShapedMF) - Smooth Z-curve transitions
• Linear Z-shaped (LinZShapedMF) - Piecewise linear Z-curve
• Pi-shaped (PiMF) - Bell with flat top

🛠️ Training options

• SGD (Stochastic Gradient Descent) – Classic gradient-based optimization with incremental updates
• Adam – Adaptive learning rates with momentum for faster convergence
• RMSProp – Scales learning rates by recent gradient magnitudes for stable training
• PSO (Particle Swarm Optimization) – Population-based global search strategy
• Hybrid SGD + OLS – Combines gradient descent with least-squares parameter refinement
• Hybrid Adam + OLS – Integrates adaptive optimization with analytical least-squares adjustment

📚 Documentation

• Comprehensive guides, API reference, and examples: docs/ (built with MkDocs).

🧪 Testing & quality

Run the full suite (pytest + coverage):

make test

Additional targets:

• make lint — Run Ruff linting
• make docs — Build the MkDocs site locally
• make help — Show all available targets with their help messages

This project is tested on Python 3.10 | 3.11 | 3.12 | 3.13 | 3.14 across Linux, Windows and macOS.

🤝 Contributing

Issues and pull requests are welcome! Please open a discussion if you’d like to propose larger changes. See the docs/guide section for architecture notes and examples.

📄 License

Distributed under the MIT License. See LICENSE for details.

📚 References

1. Jang, J. S. (1993). ANFIS: adaptive-network-based fuzzy inference system. IEEE transactions on systems, man, and cybernetics, 23(3), 665-685. https://doi.org/10.1109/21.256541