Analysis of database maintenance data using graphical and statistical techniques

In-service data for a large database system at the Rolls-Royce marine business site was analysed to investigate whether easy to obtain metrics from an information system can be used to predict likely future maintenance requirements. Graphical methods were compared and contrasted with more rigorous statistical techniques. It was found that the graphical methods could give a useful insight of possible trends and also helped identify and track the influence of external shocks to the system. The application of regression analysis was less successful with no statistically significant relationships detected. It is concluded that the flexibility and adaptability of the graphical methods are likely to be more useful to project managers.