An information system containing a set of objects is regarded. Each object has a number of attributes with attribute values related to it. The attributes are the same for all objects, but the attribute values may differ. An information system is thus more or less the same as a relational database.
An example of a decision system is shown in Table
. As one
could expect, it is a two-dimensional data table. The rows represent
objects, while the columns represent attribute values belonging to
these objects.
| studies | education | works | income | |
| 1 | no | good | yes | high |
| 2 | no | good | yes | high |
| 3 | yes | good | yes | none |
| 4 | no | poor | no | low |
| 5 | no | poor | no | medium |
In this IS there are 5 persons (objects) with attributes reflecting each persons situation of life. Assume the intention is to discover rules predicting what degree of income a person gets, depending on attributes describing him or her. The attribute income is therefore selected as a decision attribute (or dependent attribute). The rest of the attributes, studies, education, and works are then the condition attributes (independent attributes). This situation with only one decision attribute is by far the most common, and will be the main focus of this report.