Project 06-03-02: FP-Growth and Pattern Types¶
Overview¶
This project implements the FP-Growth algorithm for efficient frequent pattern mining without candidate generation. It also demonstrates maximal and closed frequent itemsets as compact representations of the full frequent itemset space.
Concepts Covered¶
- FP-Tree: compressed transaction representation as a prefix tree
- Header Table: links to all occurrences of each frequent item in the tree
- FP-Growth: recursive mining via conditional pattern bases and conditional FP-trees
- Maximal Itemsets: frequent itemsets with no frequent superset
- Closed Itemsets: frequent itemsets with no superset of equal support
Dataset¶
The Friends Cuisine dataset: 10 transactions with items from {Indian, Mediterranean, Oriental, Arabic, FastFood}. Used with min_support=3.
Files¶
fp_growth.py— complete FP-tree and FP-growth implementationproject_README.md— this file
How to Run¶
No external libraries required. Pure Python standard library.Expected Output¶
- ASCII visualization of the FP-tree
- All frequent itemsets mined via FP-Growth
- Maximal frequent itemsets
- Closed frequent itemsets
- Comparison with Apriori results