Relational Feature

See: Relational Attribute, Predictor Feature, Static Relational Attribute, Dynamic Relational Feature, Static Feature.

References

(Neville & Jensen, 2000) ⇒ Jennifer Neville, and David Jensen. (2000). “Iterative Classification in Relational Data.” In: Proceedings of the Workshop on Statistical Relational Learning.
- For example, in a data set we describe below, a relational data structure represents companies, their subsidiaries, corporate stockholders, officers and board members. Companies are linked indirectly through stockholders and through people serving simultaneously on several boards (see figure 1). Such an interlocking structure allows the creation of both intrinsic and relational** attributes. Intrinsic attributes record characteristics of objects in isolation — for example, company type or officer salary. Relational attributes summarize characteristics of one or more related objects — for example, a company’s number of subsidiaries or the maximum salary of any board member.