To perform thorough data cleansing it is necessary to first begin with data analysis and profiling. Many public pension systems tackle data cleansing one issue at a time. As they run into a data issue, that issue is addressed. We’ve seen clients’ use this approach for several years, however, until a comprehensive data analysis and profiling has been performed, data quality issues will continue to "crawl out of the woodwork". In addition, due to the complex relationships found in public pension data, sometimes certain data quality issues may not be identified until some data cleansing has occurred. For example, a service credit problem may not be known until wage reporting cleansing has taken place. The reports produced during analysis and profiling become the input into the data cleansing process.
Typically it is the responsibility of the client to actually clean the data. This may involve fixing data through automated and/or manual processes. In some cases extensive research may be needed to determine the solution and in some cases proxy data may need to be generated. ICON staff will assist clients while they cleanse data and can advise on potential solutions and options.