1. A construction company has a long history of implementing computer-based information systems to support the work of its sites. It uses a patented structured system analysis and design method. It was decided to create a computer model of the real estate market. This is an attempt, for example, to calculate the effect of interest rate changes on house values. There is some concern that the usual methodology used for IS development would not be suitable for the new project.
(a) Why is there concern and what alternative approaches should be considered?
In traditional business systems analysis, there is usually an existing office procedure that managers somehow hope to streamline. There are those who understand the details of the tasks currently being performed. They can assist the analyst in defining precise requirements to be met. In short, there is a well-defined administrative equivalent of the application to be built.
With a computer model of the real estate market, there is probably no equivalent of a well-defined administrative model. The analysts will therefore have to find ideas from various sources about how such a model works. Experts, such as experienced brokers, may be able to comment, but these are likely fragmentary rules of thumb. The academic literature on economic modeling could be studied. Ultimately, the analysts will likely have to resort to building a prototype of the model. Then they will test this to see if it behaves in a similar way to the real estate market. There is always a chance that the real estate market will behave too complex and inconsistent to accurately replicate a computer model.
All this suggests that the ‘traditional’ approach to IS development based on a waterfall framework would be inappropriate. A more iterative approach to prototyping is sensible.
(b) Prepare a system development plan that illustrates the application of your preferred methodology for this project.
- Define objectives: What does the management of the organization want from the model?
- View existing software solutions and the literature on the subject. Has anyone already done this?
- Consult experts. JAD (joint application development) sessions are mentioned in the text as a possible approach. There are many other techniques for getting relevant knowledge from domain experts that need to be considered.
- Based on the above, make a first prototype design.
- Build prototype.
- Test it, probably with real data from the recent past; you can compare the results of the model with what actually happened.
- Analyze the causes of discrepancies; change the model and run it again.
- Stop this cycle when you run out of time or get no significant improvements.
- Do a general validation of how accurate the model is, how easy it is to use.