Abstract

Telecommunication monitoring data requires the automation of data analysis workflows. A data mining tool provides data workflow management systems to process and perform analysis tasks. This paper presents an evaluation of two example data mining tools following the principles of design of experiment (DOE) to run forecasting and clustering workflows for telecom monitoring data. We conduct both quantitative and qualitative evaluation on datasets collected from a trial mobile network. The datasets consist of 1 month, six months, one year and two years of time frames that provide the average number of connected users per cell on base stations. The observations from this evaluation provide insights of each data mining tool in the context of data analysis workflows. This documented design of experiment will further facilitate replicating this evaluation study and evaluate other data mining tools.