Efficient and interpretable maximal frequent fuzzy pattern mining with multi phase pruning and ternary search.

Area of Science:

• Data Mining
• Fuzzy Set Theory
• Pattern Recognition

Background:

• Exponential growth of quantitative data necessitates efficient pattern mining.
• Traditional fuzzy frequent pattern mining faces pattern explosion and redundancy issues in dense datasets.
• Need for interpretable patterns in uncertain numerical data.

Purpose of the Study:

• Introduce a novel Maximal Frequent Fuzzy Pattern Mining (MFPM) framework.
• Address limitations of traditional fuzzy frequent pattern mining, including pattern explosion and redundancy.
• Enhance efficiency and effectiveness in mining quantitative data with numerical uncertainty.

Main Methods:

• Integration of fuzzy set theory with maximal pattern representation.
• Multi-phase approach with aggressive pruning strategies (maximum cardinality selection, early termination).
• Ternary search algorithm for identifying longest patterns and Anti-Apriori strategy with superset pruning for non-redundant patterns.

Main Results:

• Achieved up to 94.97% pattern reduction compared to traditional FTDA algorithms.
• Improved computational efficiency by over 65% in low-support scenarios.
• Demonstrated accelerated discovery in dense regimes and permissive supports.

Conclusions:

• MFPM framework generates concise, semantically interpretable patterns.
• Facilitates informed decision-making in diverse domains like healthcare analytics and business intelligence.
• Effectively handles numerical uncertainty while maintaining interpretability and reducing redundancy.