A study on Fault Identification in Continuous Integration Pipelines using Machine Learning
This study aims to develop a machine-learning model that performs fault classification on the failed test cases in the CI. We have used automated reruns of test cases with different radio software to collect a large dataset. A large dataset was be collected by using CI test results from different types of radios(example 4,8,64 ports) as inputs, and then use machine learning techniques to analyze the data and determine whether the cause of failure was related to the test environment or the software. Additionally, the model is used to investigate whether it’s possible to correlate or cluster some data to get more insights on the failure for future use; this would involve identifying the specific type of hardware that was the primary factor in the test environment, for a software failure, which framework version or software version was responsible for the failure. This thesis aims to find resource-efficient implementations of the selected algorithms and investigate the choices of the selected machine learning algorithms. The focus of this thesis is limited to state-of-the-practice deployments of machine learning algorithms adapted to the RUX way of working.