
Metrics Driven DQM - A Technical Paper
Posted by: Tata Consultancy Services
This paper is an outcome of TCS's experience in the area of data analysis by providing measurable gauges to the business for quantifying data quality.
Overview

Information has always been core of any business system and has been utilized for business optimization and growth ever since. Inappropriate information can not only lead to loss in productivity and significant cost in performing routine operations, but can even lead to losing competitive advantage in the market. It has been widely accepted that quality information is as much an asset just as people, processes and technology. Hence DQM initiatives have become backbone of all Information Management initiatives, may it be for MDM, BI, Data Integration, etc.?Quality? depends on perception and generally perception is driven by usage and requirement. Each business is governed by several sets of rules called business rules, these rules are used to run the business. Business events are tracked in the systems through various transactions. Analysis of business performance can be driven through tracking of certain Key Performance Indicators (KPI) which is derived through certain computations and aggregations. Therefore the quality of analysis is directly proportional to the quality of data captured at the operational systems. Such operational data is said to be clean if it is sufficient, consistent, accurate, latest and non-redundant. The core of data quality management process is in identifying mutually exclusive and collective exhaustive set of business rules that characterize quality data. Beyond that, the process should also be able to quantify the correctness of data based on the rules.
Features
The following acronyms have been used in this document:
Term | Explanation |
---|---|
BI | Business Intelligence |
CRR | Cleansing Rules Repository |
DQ | Data Quality |
DQA | Data Quality Analysis |
DQAR | Data Quality Analysis Report |
DQI | Data Quality Index |
DQM | Data Quality Management |
ETL | Extract, Transform, Load |
LOV | List of Values |
LTV | Length, Type, Value |
LUCAS | Latency, Uniqueness, Consistency, Accuracy, Sufficiency |
MDM | Master Data Management |