usecase: Generate and present rules

public usecase: Generate and present rules
Author: Petra Vanícková
Project: Phase: 1.0; Status: Proposed; Version: 1.0; Complexity: 1
Dates: Created: 6.12.2005 16:29:27; Modified: 10.12.2005 1:56:36;
Flags: Active: false; IsRoot: false; IsLeaf: false;
Extension Points:
UUID: {7C1B42A1-075C-4075-A2DF-77D011D331C2}
 

Goto: Scenarios

See also: Test halting conditions, Generate rules

Appears in: AD Market Basket Analysis, UC Main actions

Connections
 
Generate and present rules Scenarios
Scenario Type Detail
Basic Basic Path This part of program is responsible for rule generation. The most significant procedures are GenerateSubsets and CountConfidence.
Rules are generated from found large sets. Each such large set is recursively split by GenerateSubsets in two subsets A,B.Confidence of rule A->B is counted and compared with defined minimal confidence (represented by variable MinConf). Confidence A->B is defined as P(B|A)= # of rows containing both set B and A /# rows containing A.
Rule which confidence exceeds minimal confidence is inserted to the result and finally shown in result grid called GResult.