Temporal Data Mining

168    ◾    Temporal Data Mining

e experimental results show that the algorithm has good predictive

power and creates rules that have high interestingness.

5.3.2 Meta-Mining of Temporal Data Sets

[Abr99] addresses the issue of discovering metarules, that is, rules describ-

ing rules in large temporal data sets. Relationships between rules can give

unique insights about the characteristics of a system. Given a rule set R =

{r

1

, r

2

,…r

n

}, then a meta rule between two rules ri and rj, Mij, describes the

relationship between the two rules. e process for the discovery of the

metamining rules can be described as follows:

Rule set generation• . e rules describing the data set are generated

and separated in hierarchies, or rule set layers.

Input ge ...