Data Profiling
|
The first step in improving the quality of an enterprise’s data is to “profile†or evaluate that data. Sophisticated, yet easy to use, The data profiler is an advanced UI-based system that does not require an understanding of database engines and file structures. Business analysts or other non-technical personnel can define a set of indicators for each data element that needs to be analyzed or monitored. These indicators can range from simple or advanced statistics, to text strings, analysis, including summary data and statistical distributions of records. By reviewing the metrics on a regular basis, and following their evolution and trend, a company can follow the evolution (improvement or degradation) of the quality of its data.
|
Data Cleansing
|
Once the problem areas are identified, the data must be corrected. All data goes through a "data quality firewall" and records with missing values; values that are improperly formatted or do not match other values in the record in other data sources; duplicates; duplicates with synonyms; even simple typos -all need to be brought into alignment. This is done by cross checking against other databases and reference data.
|
Data Enrichment
|
Data Enrichment provides value-add information to the data. The variety of this information is limitless - it can include incorporating a company’s Dun & Bradstreet information or a consumer's credit score, getting the longitude and latitude of an address to help plan delivery routes, or collecting census data to target demographics or income categories. |