Mining RADIUS data : how to detect movement patterns?

• Author(s): Santos, Maribel Yasmina
• Date: 2012
• Persistent ID: http://hdl.handle.net/1822/19190
• Origin: RepositóriUM - Universidade do Minho
• Subject(s): Data mining; Radius data; Association rules; Movement patterns
• Document Type: Conference Object
• Language: English

Description

This paper presents some of the challenges that arise in the analysis of data associated to RADIUS logs, looking at movement patterns that can be identified using data mining algorithms. When talking about movement, space is inherent to the places visited by groups of individuals. However, no absolute reference to a geographic position exists in these data. The position of the Wi-Fi nodes gives some semantic about the places where people are, but does not allow the verification if two nodes are, for example, near each other. This paper describes the work developed so far in the process of analysis of such complex data set. The steps of the knowledge discovery in databases process were applied in order to verify if traditional data mining algorithms can be used in the analysis of RADIUS logs to identify places that are visited in the same sequence by several users. The obtained results show that the association rules technique can be used to identify places that are visited in a similar way by groups of individuals.