Data Quality: A Prerequisite for Succesful Data Warehouse Implementation

Abstract

Building a data warehouse for a large decentralized university such as the University of Ljubljana is an attractive challenge, but also a risky and demanding task. Experience has shown that projects attempting to integrate data are especially vulnerable to data quality issues. Therefore, before embarking on a data warehouse initiative a thorough quality assessment of the source data is necessary. We describe how the assessment criteria based on the Total Quality data Management Methodology were adapted to our specific needs and used to determine the quality of student records data at two member institutions, viz. the Faculty of Computer and Information Science, and the Faculty of Electrical Engineering. The most important results of the assessment are described and proposals are given for further activities. The assessment has shown that the student records data at the Faculty of Computer and Information Science and Faculty of Electrical Engineering are good enough to be used as source for the global warehouse at the university level after some data cleansing takes place. Additionally, special attention must be devoted to the integration of such data that are replicated at many individual departments (viz. employees, subjects taught, and students). Therefore, we propose that a unique coding scheme for all employees, students, and subjects taught be defined in the first step of the data warehouse design, and an ongoing data quality management process is established clearly defining the roles and responsibilities of all personnel involved.