Data problems in decision support systems

A major problem with the usefulness of decision support systems (DSS) or executive information systems (EIS) occurs when the data used by the system is not accurate. When inaccurate data is used as input to a DSS, the resulting output may seem illogical to management. As a result, the management views the EIS as defective. To prevent this problem, it is wise for the systems analyst who designs the system to evaluate the accuracy of the source data and to understand the accounting methods used to generate the data. This paper looks at accounting practices, outliers and other practices that may produce inaccurate or misrepresentative data. Often, the problem is that the data is atypical because of an impulse from some normal pattern. The impulse is assumed to be nonrecursive and not to follow a discernible pattern over time. When an impulse is induced by inventory adjustments, strikes, price adjustments, and mergers, it is necessary for the EIS to be designed to accommodate the effect of the atypical data. Decision models which ignore unexpected situations or use inaccurate data are of limited value, for in practice management decision models are evaluated on the basis of infrequency of failures rather than frequency of successes.<>