Assessing the source code quality of software objectively requires a well-defined model. Due to the distinct nature of each and every project, the definition of such a model is specific to the underlying type of paradigms used. A definer can pick metrics from standard norms to define measurements for qualitative assessment. Software projects develop over time and a wide variety of re-factorings is applied tothe code which makes the process temporal. In this thesis the temporal model was enhanced using methods known from financial markets and further evaluated using artificial neural networks with the goal of improving the prediction precision by learning from more detailed patterns. Subject to research was also if the combination of technical analysis and machine learning is viable and how to blend them. An in-depth selection of applicable instruments and algorithms and extensive experiments were run to approximate answers. It was found that enhanced patterns are of value for further processing by neural networks. Technical analysis however was not able to improve the results, although it is assumed that it can for an appropriately sizedproblem set.
| Identifer | oai:union.ndltd.org:UPSALLA1/oai:DiVA.org:lnu-51864 |
| Date | January 2015 |
| Creators | Hönel, Sebastian |
| Publisher | Linnéuniversitetet, Institutionen för datavetenskap (DV) |
| Source Sets | DiVA Archive at Upsalla University |
| Language | English |
| Detected Language | English |
| Type | Student thesis, info:eu-repo/semantics/bachelorThesis, text |
| Format | application/pdf |
| Rights | info:eu-repo/semantics/openAccess |
Page generated in 0.0017 seconds