Source & licenses.

Application files

Buildless HTML, CSS, JavaScript and a shared Python experiment engine. Serve the dist folder with any static HTTP server.

index.html · style.css · app.js · explorer.js · worker.js · engine.py

exFuzzy 3.3.1

The official, unmodified Python wheel is included with the app. exFuzzy is licensed under AGPL-3.0-or-later. The browser uses its FuzzyRulesClassifier, native fuzzy partitions, rule mining and genetic rule selection.

Version 3.3.1 source · Download exFuzzy source · Official wheel

PyMoo in the browser

The genetic learners use PyMoo 0.6.1.6 (Apache-2.0), vendored as a pure-Python wheel built from the official package without its optional Cython speed-ups, together with its pure-Python dependencies: autograd (MIT), cma (BSD-3-Clause), alive-progress (MIT), about-time (MIT), grapheme (MIT), Deprecated (MIT) and wrapt (BSD-2-Clause). moocore is a small stub that only lets PyMoo import; exFuzzy's single-objective search never uses it. All wheels are in vendor/.

Runtime and datasets

The Python runtime is Pyodide 0.29.5 (MPL-2.0) with NumPy, pandas, SciPy and scikit-learn (BSD-3-Clause), micropip (MPL-2.0), joblib (BSD-3-Clause), threadpoolctl (BSD-3-Clause), python-dateutil (Apache-2.0 / BSD-3-Clause), pytz (MIT) and six (MIT) from the official Pyodide package distribution, mirrored in pyodide/ so the app loads them from this site. DM Sans and Manrope are served by Google Fonts. Experiment data is never transmitted.

Iris is scikit-learn's Iris dataset. The material screening dataset is synthetic, generated with seed 2026, and is intended only to demonstrate mixed numeric and categorical inputs.