liveBook · Manning

Conditional pattern base

As explained in Data Without Labels

A conditional pattern base is the collection of prefix paths and associated counts leading to a specified item in a frequent pattern tree.

A conditional pattern base collects the prefix paths and counts leading to a selected item in a frequent pattern tree. It supports extracting patterns for that item. section 4.6

Construct one by tracing every path from the root to the selected item, removing the NULL root, and recording each prefix with the item's count. Aggregate repeated prefixes, discard counts below the minimum support, form the conditional F-P tree, then append the original item. section 4.6

Cheese

Cheese appears through Milk-Cheese and Milk-Apples-Cheese, producing {{Milk:1}, {Milk-Apples:1}}. Aggregation gives Milk:2 and Apples:1; with minimum support count 2, retain {Milk:2} and generate {Milk-Cheese:2}. section 4.6

How Data Without Labels covers it

How deep it goes
Explains path collection, count aggregation, support pruning, and pattern generation, but does not detail library implementation.
Examples
Uses Cheese paths, Milk and Apples counts, and a tabular F-P algorithm dataset.
What it assumes
Assumes familiarity with transaction item sets, support counts, and frequent pattern trees.
What you can do after
Apply conditional pattern bases to derive conditional F-P trees and frequent patterns for selected items.
How it connects
Links frequent pattern trees, conditional F-P trees, minimum support, and association rules.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Data Without Labels. The section links point to the parts of the book it draws on.