7/8/2002

Student Name: Yu Wang

Committee:
Dr. Jianhua Chen
Dr. Donald H.Kraft
Dr. J.Bush Jones

Date: Friday, July 12, 2002
Time: 9:30 am
Location: 297 Coates Hall

Title: A Web Based Data Mining Application for Demographic Database

Abstract:
A web based client-server application for determining people's potential earning based on demographic profile was implemented using Java Servlet and two data mining methods: Decision Tree and Bagging. The database has 30,162 training data and 15,060 testing data, and is characterized by 13 attributes including age, education, work class, occupation, sex, capital gain/loss, etc. The instance space is over 1.3 billion.

This project implemented three-tier client/server architecture with web browser as client, Java Servlet as a middle tier and database as the third tier connected through SQL server 2000 JDBC driver. User can query, insert, delete, update the adult database on line based on their permission.

For the data-mining component of the project, two different but related methods are implemented, ID3 decision tree algorithm and bagging algorithm. The comparison between these two algorithms is analyzed according to the result.