Data quality and data cleaning: an overview

Data quality is a serious concern in any data-driven enterprise, often creating misleading findings during data mining, and causing process disruptions in operational databases. The manifestations of data quality problems can be very expensive- "losing" customers, "misplacing" billions of dollars worth of equipment, misallocated resources due to glitched forecasts, and so on. Solving data quality problems typically requires a very large investment of time and energy -- often 80% to 90% of a data analysis project is spent in making the data reliable enough that the results can be trusted.In this tutorial, we present a multi disciplinary approach to data quality problems. We start by discussing the meaning of data quality and the sources of data quality problems. We show how these problems can be addressed by a multidisciplinary approach, combining techniques from management science, statistics, database research, and metadata management. Next, we present an updated definition of data quality metrics, and illustrate their application with a case study. We conclude with a survey of recent database research that is relevant to data quality problems, and suggest directions for future research.