mining negative rules using grd

Post on 30-Dec-2015

42 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

DESCRIPTION

Mining negative rules using GRD. D. R. Thiruvady and G. I. Webb PAKDD 2004. Outline. Introduction OPUS GRD Conclusion. Introduction. Association rule A ==> B (A is antecedent ,B is consequent) Negative Rules Either antecedent or consequent or both are negated. - PowerPoint PPT Presentation

TRANSCRIPT

2006/10/25 1

Mining negative rules using GRD

D. R. Thiruvady and G. I. Webb

PAKDD 2004

2006/10/25 2

Outline

• Introduction

• OPUS

• GRD

• Conclusion

2006/10/25 3

Introduction

• Association rule – A ==> B (A is antecedent ,B is consequent)

• Negative Rules– Either antecedent or consequent or both are negated

A B A B

A B

2006/10/25 4

Optimized Pruning for Unordered Search (OPUS)

2006/10/25 5

Generate Rule Discovery (GRD)

• Extends OPUS by remove the requirement that consequent be single variable

• K number of rules replace minimum support

2006/10/25 6

GRD

• Four measures with respect to a rule XY

2006/10/25 7

GRD

• Symbol

2006/10/25 8

Properties

2006/10/25 9

Properties

2006/10/25 10

Properties

2006/10/25 11

Properties

2006/10/25 12

GRD

• Symbol– CurrentLHS: Init ψ– AvailableLHS: Init Antecedent– AvailableRHS: Init Consequent

• Function– Insolution(ac): rule ac in solution– Proven(X): pruning rules provided to the alg

orithm prove the proposition X

2006/10/25 13

GRD algorithmPrune 1

Prune 2

Prune 3,4,5

2006/10/25 14

GRD algorithm2Prune 6

2006/10/25 15

Pruning

2006/10/25 16

Pruning

2006/10/25 17

Pruning

2006/10/25 18

Update Constraints

2006/10/25 19

Negative with GRD

support(A & B) support(A B) support(A & B)x x

support(A B) support(A) support(A B)

2006/10/25 20

Conclusion

• Disadvantage– Search space too large

top related