uy

Based on https://github.com/pbharrin/machinelearninginaction/blob/master/Ch12/fpGrowth.py

To install: pip install uy

Description

The uy package implements the FP-Growth algorithm for frequent itemset mining, avoiding the costly generation of candidate sets involved in algorithms like Apriori. This implementation includes functions to construct the FP-tree, update it, and mine the frequent itemsets from it. The package is designed to efficiently find frequent itemsets in a given dataset, which is crucial for tasks such as market basket analysis, association rule learning, and anomaly detection.

Main Components

• treeNode: A class representing a node in the FP-tree. Each node contains links to parent and child nodes, a count of occurrences, and methods to manage the node's data.
• createTree: A function to build the FP-tree from the dataset. It also constructs a header table that helps in tree traversal.
• updateTree: Used to add items to the FP-tree during its construction.
• mineTree: Once the FP-tree is constructed, this function is used to mine the frequent itemsets from the tree using the header table.
• loadSimpDat: A utility function to load a simple example dataset.
• createInitSet: Converts a list of transactions into a dictionary format expected by createTree.

Usage Examples

Loading Data and Creating Initial Set

from uy import loadSimpDat, createInitSet

# Load example data
simpDat = loadSimpDat()

# Create initial set from data
initSet = createInitSet(simpDat)

Building the FP-Tree

from uy import createTree

# Minimum support
minSup = 3

# Create FP-tree and header table
myFPtree, myHeaderTab = createTree(initSet, minSup)

Mining Frequent Itemsets

from uy import mineTree

# List to hold the mined frequent itemsets
freqItems = []

# Mine the tree
mineTree(myFPtree, myHeaderTab, minSup, set([]), freqItems)

# Print the frequent itemsets
print(freqItems)

Documentation

Class: treeNode

• init(self, nameValue, numOccur, parentNode): Initialize a new tree node.
• inc(self, numOccur): Increment the count of occurrences for the node.
• disp(self, ind=1): Display the subtree rooted at this node.
• str(self, ind=1): Return a string representation of the subtree rooted at this node.
• repr(self, ind=1): Return the string representation for interactive environments.

Function: createTree

• createTree(dataSet, minSup=1): Create the FP-tree from the dataset. It returns the root of the FP-tree and the header table.

Function: updateTree

• updateTree(items, inTree, headerTable, count): Update the FP-tree with given items.

Function: mineTree

• mineTree(inTree, headerTable, minSup, preFix, freqItemList): Mine the FP-tree to find frequent itemsets that meet the minimum support.

Function: loadSimpDat

• loadSimpDat(): Load a simple hardcoded dataset for demonstration purposes.

Function: createInitSet

• createInitSet(dataSet): Convert dataset into a format suitable for the FP-tree construction.

By using these functions and classes, users can perform efficient frequent itemset mining in various datasets, which is a foundational technique in many data mining applications.