A rule-based approach to variance analysis

This research presents a more complete model to address a significant accounting issue, variance analysis. Past research studies have simplified the variance analysis process. By relying on quantitative parameters to implement models, researchers have omitted important information flows and cues. Artificial intelligence technology was used in this study to incorporate many of these variables. The expert system, created from written and human sources of expertise, uses a knowledge base of rules, parameters, and control strategies to decide whether a variance is worthy of investigation. The system analyzes material and labor variances in an attempt to test the possible improvement of variance investigation using a knowledge-based methodology.

Validation procedures demonstrated the system's ability to process test cases and generate desired output. Therefore, a rule-based approach to variance analysis proved successful at the initial prototype stage. System performance was also compared to a traditional variance investigation model, X̄ control chart. Similarities or differences between model results were due to varying model emphasis. A control chart focuses on the statistical attributes of a variable. The expert system was specifically designed to analyze a deviation from a management desired cost level. This variable requires more analysis than simple statistical limits testing.