|
Activity Number:
|
522
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 7, 2008 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biopharmaceutical Section
|
| Abstract - #302471 |
|
Title:
|
Using the Z-Test To Sort Association Rules in Large Spontaneous Reporting Databases
|
|
Author(s):
|
Chandra S. Thames*+
|
|
Companies:
|
Bracy Analytics Inc.
|
|
Address:
|
2300 Lakeview Parkway , Alpharetta, GA, 30004,
|
|
Keywords:
|
association rules ; data mining ; vaccine safety ; drug safety
|
|
Abstract:
|
For many years the focus in 'mining' association rules in large databases has been on the development of more efficient algorithms for finding such rules. However, in recent years research has also focused on statistical techniques for sorting the rules. We recommend sorting the rules by the z-test of the lift instead of the adjusted lift by the Gamma Poisson Shrinkage (GPS) estimator. The GPS estimator is currently the only method used for searching the FDA's Spontaneous Reporting System database and the CDC's Vaccine Adverse Events Spontaneous Reporting System for 'interesting' drug-event combinations. The mean and variance of the lift are derived in order to perform the z-test. The GPS was developed to 'shrink' the lift, adjusting for small counts. However, those rules with low expected counts still dominate the top after applying GPS.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2008 program |
