Data quality improvement can be done in every managing process (collecting process, analysis process and displaying process). By observing the errors in processing data, we can avoid or minimize those errors so that improve the data quality. Data with good quality is valid, reliable and representative. It is valid if the data reflects the real characteristic of the object. It is reliable if the characteristic has consistent dependent or stabile so that accurate in explaining the object. It is representative if the data can represent the population. Besides that, good data also complete, accurate and up to date, easy and fast to be accessed.
There are six steps of data quality improvement. First, define the problem in process context. This begins with the settlement or specification of which system is involved so that the efforts are focused on process, not output. Second is identification and process documentation. Flowchart is common tool that is used to describe the process. The identified process must be documented well so that it can be used as information continuously. Third, performance measurement which is to quantify how good or how bad is the current system. Basically, there are three levels of performance measurements that are process, output and outcome. These measurements define the activity, variable and operation from the working process itself.
The fourth step in data quality improvement is to understand why a problem in process context can happen. If there is no data then it will be difficult to understand why a system is running that way so that the performance is not as expected. A problem is the deviation that happens between the expected performance (target) and actual performance (result). Fifth, develop and test the ideas. The chosen ideas can be effective if it is tested before implemented to avoid the failure in the process. The last step is solution implementation and evaluation.