|Keywords:||Network support data; Risk level; Machine learning; Engineering and Technology; Electrical Engineering, Electronic Engineering, Information Engineering; Computer Systems; Teknik och teknologier; Elektroteknik och elektronik; Datorsystem; Computer science; Datavetenskap|
|Full text PDF:||http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-128656|
Internet Service Providers gather vast amounts of data in the form of trouble tickets created from connectivity related issues. This data is often stored and seldom used for proactive purposes. This thesis explores the feasibility of finding correlations in network support data through the use of data mining activities. Correlations such as these could be used for improving troubleshooting or staffing related activities. The approach uses the data mining methodology CRISP-DM to investigate typical data mining operations from the perspective of a Network Operation Center. The results show that correlations between the solving time and other ticket related attributes do exist and that support data could be used for the activities mentioned. The results also show that it exists a lot of room for improvement when it comes to data mining activities in network support data.