Decision Support as a Web Service
This namespace was used to support an MSc research project evaluating the feasibility of providing browser based Decision Support functionality as a Web Service.
Of particular interest to the author were query latencies and server loading metrics which directly affect user interactivity, response and his ability to remain focused in researching facts supporting a hypothesis for decision.
The project is principally an OLAP-based decision support tool providing a rich graphical user interface supported by drag-and-drop dimensional views and measures onto cross-linked pivot chart and pivot table work areas.
Emphasis was placed on an easy user interface to complex multi-dimensional graphical representation. Visualization is facilitated by encouraging dropping measures and dimensions onto the chart rather than fixation on pivot table numbers.
Highly structured sales and service automotive dealership domain information spanning 13 years was used in the construction of the experimental model. The vehicle service history hypercube contains 13 logical fact dimensions and aggregates 14 measures while the vehicle sales history hypercube contains 11 fact dimensional views and a unit measure.
The project solution is deterministic and serves best as an analytical tool for detecting patterns visually and aiding in locating problem scenarios, particularly as it relates to the hierarchical drill-down and filtering capabilities provisioned in the cube geometries.
The solution makes no attempt to apply cognitive, heuristic, predictive or statistical evaluation on the information, however a skilled manager will find the tool effective in discovering a wealth of hidden knowledge.
The dissertation was graded A.
|© 2005 - Gus Constan - BEng, MSc|