Alice in the Fuzzy Technologies laboratory, introducing FuzzyRoutines

Python · Fuzzy mathematics · Open source

Turn measurements into interpretable grades with membership functions, fuzzy sets and linguistic scales. Build models for expert systems, risk assessment and engineering decisions.

CPython 3.13–3.14 · Typed scalar API · No mandatory NumPy dependency · Apache-2.0

Try an example Explore the API

Technologies · Knowledge · Science

Quick start

How well does a room temperature of 24 °C fit the idea of “comfortable”? Define full comfort at 22 °C, decreasing to zero at 16 and 28 °C, then evaluate the measurement.

1 · Install

Use Python 3.13 or 3.14 in your project environment.

This example uses the 2.0 API. Install the matching version from PyPI; historical 1.x releases do not provide this API.

Terminal
python -m pip install fuzzyroutines==2.0.0

2 · Run the example

Save this as comfort.py and run python comfort.py.

Python · comfort.py
from fuzzyroutines import (
    ContinuousUniverse, ScalarFuzzySet, Triangle,
)

universe = ContinuousUniverse(
    0, 40, leftClosed=True, rightClosed=True,
)
comfort = ScalarFuzzySet(universe, Triangle(16, 22, 28))
temperature = 24
grade = comfort.Membership(temperature)

print(f"Temperature: {temperature} °C")
print(f"Comfort membership: {grade:.3f}")

3 · Read the result

Expected output
Temperature: 24 °C
Comfort membership: 0.667

The temperature is on the descending side of the triangle: (28 − 24) / (28 − 22) = 2/3. A grade of 0.667 describes partial membership in the comfort set, not a probability.

These thresholds are illustrative model choices. Add other terms to distinguish comfortable, warm and hot conditions in your application.

Continue with a two-term scale →
Comfort triangle and Warm shoulder, with the 24 °C observation marked
The full tutorial adds a Warm term with membership 0.125 at 24 °C.

Universal Fuzzy Scale

Historical model · Modern API

Five levels for a normalized measurement

Min · Low · Med · High · Max

Reconstruct the historical scale with its original coefficients and compare it with the modern implementation. The guide includes membership curves, boundary cases and selection rules.

Use linguistic levels to describe normalized risk scores, security findings or control effectiveness. Define what a higher score means for your model; fuzzy membership is not an incident probability.

Explore the scale and graphs →

Build your fuzzy model

Define

Membership and sets

Eight membership families, explicit universes and immutable scalar fuzzy sets.

Choose a curve →
Combine

Fuzzy operations

Norms, negations, intersection, union and directed difference with explicit policies.

Worked scenarios →
Interpret

Scales and centroids

Linguistic classification, alpha-cuts and centroid computation with exact or numerical evidence.

API reference →

Documentation

Nine worked scenarios, reproducible graphs and a reference for the public API.

Upgrading from 1.0.3 → Changelog → Implementation status →
Scope and compatibility

Implemented in 2.0: the modern scalar API and historical compatibility facade, with explicit validation and documented numerical corrections.

Future work: additional defuzzification methods, executable convexity queries, symmetric difference and production vectorized execution. These are outside the current scalar API.