When dealing with large amounts of heterogeneous traffic data streams, how a Complex Event Processing (CEP) system, which can efficiently process and predict complex events in traffic, is set up is a crucial matter. In this paper, several issues and methods related to finding different rules that can be used to develop such a system are presented. Statistical methods to detect complex events from traffic data are first described. Two types of techniques are used to research relations between complex events: descriptive and predictive data mining. First, association rules are used to analyze data and express regularities in data. Second, decision trees and decision rules algorithms are used for the prediction of complex events. All the algorithms were tested with regards to how different social events affect the traffic system near the Stozice stadium in Ljubljana, Slovenia. The results show that methods described in this paper are feasible and can be used for developing an advanced traveler information system.