Page 97 -

64 Part 1 • SyStemS analySiS FundamentalS

Figure 3.9 Software Requirements Specific Software Features
Guidelines for evaluating
software. Performance Effectiveness Able to perform all required tasks
Able to perform all tasks desired
Well-designed display screens
Adequate capacity
Performance Efficiency Fast response time
Efficient input
Efficient output
Efficient storage of data
Efficient backup
Ease of use Satisfactory user interface
Help menus available
“Read Me” files for last-minute changes
Flexible interface
Adequate feedback
Good error recovery
Flexibility Options for input
Options for output
Usable with other software
Quality of Documentation Good organization
Adequate online tutorial
Website with FAQ
Manufacturer Support Technical support hotline
Newsletter/email
Website with downloadable product updates

Identifying, Forecasting, and Comparing Costs
and Benefits
Costs and benefits of a proposed computer system must always be considered together because
they are interrelated and often interdependent. Although a systems analyst tries to propose a
system that fulfills various information requirements, decisions to continue with the proposed
system will be based on a cost-benefit analysis, not on information requirements. In many ways,
benefits are measured by costs, which becomes apparent in the next section.

Forecasting
Systems analysts are required to predict certain key variables before submitting a proposal to the
client. To some degree, a systems analyst will rely on a what-if analysis, such as, “What if labor
costs rise only 5 percent per year for the next three years, rather than 10 percent?” The systems
analyst should realize, however, that he or she cannot rely on what-if analysis for everything if
the proposal is to be credible, meaningful, and valuable.
A systems analyst has many forecasting models available. The main condition for choos-
ing a model is the availability of historical data. If historical data are unavailable, the analyst
must turn to one of the judgment methods: estimates from the sales force, surveys to estimate
customer demand, Delphi studies (a consensus forecast developed independently by a group of
experts through a series of iterations), creation of scenarios, or historical analogies.
If historical data are available, the next differentiation between classes of techniques
involves whether the forecast is conditional or unconditional. Conditional implies that there is
an association among variables in the model or that such a causal relationship exists. Common
methods in this group include correlation, regression, leading indicators, econometrics, and
input/output models.
Unconditional forecasting means the analyst isn’t required to find or identify any causal rela-
tionships. Consequently, systems analysts find that these methods are low-cost, easy-to-implement