In business life (but naturally in personal matters, too), many times you have to face the problem of selecting the best one or some from a large set of candidates. You know which are the interesting attributes, you can even compare some of them - but in most cases the preferences are conflicting, all the alternatives are good in some and bad from other viewpoints. To derive a correct decision, you should have to calculate a "fitness value" for all the alternatives and overview the results.

This process is supported by the Decision Support Module (which was formerly a standalone application, DecArt, but was integrated into DataScope as a module).

Working with this module consists of several steps. First you have to select the attributes (decision criteria) that you want to take into account when deriving the decision, then you have to assign weight (importance) to each. This would not be easy if you had to assign a number to each, as criteria mostly come from far different areas, for example an applicant for a position has personal, professional, etc. values. Inside a category, you can compare members (language knowledge, typing speed, etc.), you can also compare the groups (for example "professional values" is more important than "personal values"), but cannot compare across categories. In DataScope, you can build a hierarchical structure from the groups and criteria, and compare only the members of a hierarchy group.

Next, you have to specify your preferences among the possible values of a field (the higher typing rate is the better) using a sophisticated wizard.

The rest is done by the algorithm: it calculates the preference value for each alternative. Naturally, you are interested in the preference value on the top level of this structure - the top ones are the best according to your decision concept; but the algorithm calculates this value for all subtree items and ranks the alternatives according to those values also, so you can extensively survey the results.

At the same time, alternatives are also clustered by your preferences, the alternatives "more similar" from your viewpoint get into the same cluster (like "young, inexperienced, cheaper" and "older, experienced, more expensive" applicants). You can select the best cluster before checking the best alternatives.

DataScope also provides specialized charts to compare two or several alternatives and to visually display the result of the comparison.

Similarly to other DataScope algorithms, all results of this algorithm are stored as data columns and can be displayed on other charts or reused in expressions, models, and so on.

A special feature of the Decision Support Module is that it can learn the decision concept backward, from the results. If you have a set of alternatives, and you can assign your personal preference value, it can learn and build the decision concept which leads to that ranking. Thus, you can find out the important factors from a few known decision examples, and can rank all, yet unknown, alternatives based on this decision concept.

Decision Support Module Features:

Multi-criteria decision support algorithm
Handling hierarchical decision concepts
Simple, easy to use interface to build the concept
Sophisticated interactive visual evaluation of the result
Backward learning of the decision concept from a few given examples