Software is present in almost all aspects of our lives, and with more parts of life beingdriven by code, the importance of limiting bugs is critical. Studies have shown that thelonger a bug is present in a system increases the complexity of finding and handlingthe bug. Tools aimed to find bugs before they enter a system exist, but usually requiressignificant amount of data before working. This thesis aims to investigate if the stateof-the-art bug detection tool, JITLine, can be used on projects from the start. Thiswas investigated by performing cross project bug predictions, by training on an alreadyexisting system and then using the trained model on a new system. Different iterationswere done to simulate varying scenarios. It was quickly shown that a good matchbetween the project where data was gathered from and JITLine was crucial. But evenfor the best match that was evaluated, the result was significantly lower than what isto be considered a good result in the Machine Learning (ML) world. Therefor it wasconcluded that JITLine was not suited for cross project bug predictions, and therefornot suited for new projects
Identifer | oai:union.ndltd.org:UPSALLA1/oai:DiVA.org:kau-98326 |
Date | January 2024 |
Creators | Pettersson, Axel |
Publisher | Karlstads universitet, Institutionen för matematik och datavetenskap (from 2013) |
Source Sets | DiVA Archive at Upsalla University |
Language | English |
Detected Language | English |
Type | Student thesis, info:eu-repo/semantics/bachelorThesis, text |
Format | application/pdf, application/pdf |
Rights | info:eu-repo/semantics/openAccess, info:eu-repo/semantics/openAccess |
Page generated in 0.0139 seconds