Clean Data In

"Garbage in, garbage out ...," the opening sentence of "Take Your Data to the Cleaners" (January State of the Art), is somewhat ironic. The article certainly addresses what to do with "dirty" data, yet it doesn't address the "garbage in" part of the equation. If you're going to invest effort in cleaning your data, you should also control how you enter data in the first place. If your workers enter months as Jan., January, J, and 01, you clearly need to standardize input procedures. Training helps, but automating quality control might yield better results: Use formatted data fields with built-in error checking, pi ck data from pick lists, verify data with lookup tables, and post-process inputs before committing them to the database. Controlling the "in" part of the equation makes the data more immediately usable, might increase productivity, and might even revea l some properties of your data that you never suspected.

Geoff Hart
geoff-h@mtl.feric.ca