Version: 0.5 Date: Aug 01, 2025
Zuffy - Fuzzy Pattern Trees induced by Genetic Programming
An Open Source sci-kit compatible classifier.
This package is part of an academic research exercise and it is designed to help a Machine Learning developer to explore Fuzzy Pattern Trees (FPT).
FPTs are models that have high explainability and they help the researcher explore a dataset where vague or ambiguous data may be present.
This kit will analyse a dataset and fuzzify the input features to generate a collection of fuzzy sets. These sets become new features which are combined with fuzzy operators to build a pattern tree.
The package uses Genetic Programming to induce the tree and it requires the SymbolicClassifer from the open source gplearn package to support the genetic evolution process.
Contents:
- Zuffy Classifier
- Zuffy Fit Iterator
- Fuzzy Transformer
- Visuals
- Visual Color Assigments
- _fpt_operators
_complement()_concentrator()_concentrator_power()_diffuser()_diluter()_diluter_power()_fuzzy_and()_fuzzy_or()_generate_owa_operator()_generate_wa_operator()_hamacher_t_norm()_if_gte()_if_gte_else()_if_lt()_if_lt_else()_intensifier()_lukasiewicz_t_conorm()_lukasiewicz_t_norm()_maximum()_minimum()_ordered_weighted_average()_product_t_norm()_weighted_average()
- Sample Programs