fitron 0.1.3

FITRON: adaptive fuzzy-tree MCDM decision engine

FITRON

FITRON is a Python package for hybrid decision intelligence under uncertainty.

The package combines:

• fuzzy feature transformation,
• decision tree learning,
• TOPSIS multi-criteria ranking,
• memory-guided adaptive weight updates,
• interpretable ranking outputs.

Package name on PyPI: fitron
Import path: fitron

Why FITRON

Most pipelines stop at classification. FITRON extends that flow into adaptive decision ranking.

1. Learns a predictive backbone with a decision tree.
2. Ranks valid candidates with TOPSIS and learned criteria weights.
3. Adapts weights over iterations using memory and exploration.
4. Produces feature-level explanations for selected options.

Core Workflow

1. Preprocess tabular data and encode features.
2. Fuzzify numeric features into low/mid/high memberships.
3. Train decision tree model and estimate feature importances.
4. Filter valid candidates using model predictions.
5. Rank candidates with TOPSIS.
6. Update weights using adaptive exploration and memory feedback.
7. Return best option, score, and explanation.

Installation

End users

pip install fitron

From source

pip install .

Editable install (development)

pip install -e .[dev]

Environment Setup

Use a standard CPython virtual environment.

Windows:

python -m venv .venv-win
.\.venv-win\Scripts\python.exe -m pip install -U pip setuptools wheel
.\.venv-win\Scripts\python.exe -m pip install -e .[dev]

If PowerShell blocks script activation, use:

.venv-win\Scripts\activate.bat

Linux or macOS:

python3 -m venv .venv
source .venv/bin/activate
python -m pip install -U pip setuptools wheel
python -m pip install -e .[dev]

More detailed setup for Linux/macOS:

1. Confirm Python 3.10+ is available:

python3 --version

2. If venv is missing on Linux, install it first (example for Debian/Ubuntu):

sudo apt update
sudo apt install -y python3-venv

3. Create and activate the virtual environment:

python3 -m venv .venv
source .venv/bin/activate

4. Upgrade packaging tools and install FITRON in editable mode:

python -m pip install -U pip setuptools wheel
python -m pip install -e .[dev]

5. Verify installation:

python -c "from fitron import FITRONModel; print('ok')"
pytest -q

Notes:

• On some macOS setups, python3 may be installed via Homebrew and located under /opt/homebrew/bin/python3 (Apple Silicon) or /usr/local/bin/python3 (Intel).
• Use deactivate when you are done working in the virtual environment.

Quick Start

import pandas as pd
from fitron import FITRONModel

sample = pd.DataFrame(
    {
        "income": [50000, 20000, 75000, 43000],
        "risk": [0.2, 0.8, 0.3, 0.5],
        "credit_score": [710, 520, 760, 640],
        "target": [1, 0, 1, 1],
    }
)

model = FITRONModel(iterations=5, random_state=42)
result = model.fit(sample, target="target")

print("Best index:", result.best_index)
print("Best score:", result.best_score)
print("Explanation:", result.explanation)

Public API

• FITRONModel
• fit(df, target, ...)
• rank(df, target, ...)
• explain(result)
• update_memory(memory, weights, score, best_idx)

Run Tests

pytest -q

Build and Publish

Build distribution artifacts:

python -m build

Validate artifacts:

python -m twine check dist/*

Upload:

python -m twine upload --repository testpypi dist/*
python -m twine upload dist/*

Project Structure

src/pip_model/
  core/
  api.py
  pipeline.py
tests/
examples/
pyproject.toml

Authors

• zibransheikh
• hazlived

License

MIT