Data Profiling: Assessing Data Quality

Data profiling enables you to assess the quality of your source data before you use it in data integration scenarios and systems. OWB provides the Data Profile Wizard to guide you through creating a data profile, and the Data Profile Editor to configure and manage data profiles. Because data profiling is integrated with the ETL features in OWB and other data quality features, such as data rules and built-in cleansing algorithms, you can also generate data cleansing mappings and schema correction scripts. This enables you to automatically correct any inconsistencies, redundancies, and inaccuracies in both the data and metadata.

Data profiling enables you to discover many important things about your data. Some common findings include:

• A domain of valid product codes

• A range of product discounts

• Columns that hold the pattern of an e-mail address

• A one-to-many relationship between columns

• Anomalies and outliers within columns

• Relations between tables even if they are not documented in the database

To begin the process of data profiling, you first use the Data Profile Wizard to create a data profile from within the Design Center. You then use the Data Profile Editor to run data profiling on the objects contained in the data profile, and to create correction tables and mappings.