Department of Computer Science and Engineering UNIVERSITY OF GOTHENBURG CHALMERS UNIVERSITY OF TECHNOLOGY Gothenburg, Sweden 2018 Statistical Model Update Optimization in Industrial Practice

Abstract

This thesis presents a study done on optimizing machine learning model updates. The department of Quality and Functionality in a multinational telecommunication company is searching for an optimal solution to the problem of when, and how, to trigger a training cycle of a statistical model on their test execution dataset. We have investigated techniques regarding the possibilities of optimizing a statistical model update. A case-study has been conducted, using a telecommunication company as a case subject company.