Sandbox‎ > ‎Archive‎ > ‎IPT 2008-09‎ > ‎Blair's page‎ > ‎Blair's updates!‎ > ‎

2009-06-11 DSS types

posted Jun 10, 2009, 5:19 PM by Unknown user   [ updated Jun 11, 2009, 12:16 AM by Eddie Woo ]
For further details, refer to IPT: DSS Types


The kinds of analysis supported by spreadsheets include the 'drill-down analysis', the 'what-if analysis' and 'goal-seeking'. The 'drill-down analysis' is the act of obtaining more details about a solution provided by a decision support system, so that the end user may know how the solution provided by the DSS was obtained. The 'what-if analysis' is the act of making changes to data in order to observe the effects of the changes. 'Goal-seeking analsis' is the act of locating data that is required to produce a result.

An example of a problem that is effectively supported by a spreadsheet is the need to find the total resistance of the resistors in a parallel circuit. This can be done by programming a mathematical formula for the total resistance and inputting the data. As the formula does not change or rely on heuristics, the DSS does not need to make any assumptions and still be correct every time (assuming that the input data is correct).

An example of a problem that cannot be effectively supported by a problem is the need to find the most delicious ingredients to use in chicken soup. This simply cannot be calculated with a spreadsheet, as it is not a structured problem.


Relational databases can compile data from many sources (which are commonly grouped into categories by topic) in a manner which does not confuse the end user or system developer. This is highly important when there is a need to program formulae or logical processes. Relational databases are not confused for the end user because they make use of structured queries, which can be used to quickly locate data.

Expert systems

Expert systems mimic real human experts by being specialised (domain-specific) and making extreme use of heuristics to try to know what to expect in complex situations.

AI systems are difficult to design because a huge amount of data and information need to be programmed into them, including specialised information as well as what heuristics are to be used. They are difficult to maintain because real-world data is more often than not ephemeral, and hence this data needs to be constantly updated to reflect changes in the real world.