상세정보
Export to Refworks부가기능
Learning Actionable Analytics in Software Engineering
상세 프로파일
| 자료유형 | 학위논문 |
|---|---|
| 서명/저자사항 | Learning Actionable Analytics in Software Engineering. |
| 개인저자 | Krishna Prasad, Rahul. |
| 단체저자명 | North Carolina State University. |
| 발행사항 | [S.l.]: North Carolina State University., 2019. |
| 발행사항 | Ann Arbor: ProQuest Dissertations & Theses, 2019. |
| 형태사항 | 179 p. |
| 기본자료 저록 | Dissertations Abstracts International 81-05B. Dissertation Abstract International |
| ISBN | 9781392648919 |
| 학위논문주기 | Thesis (Ph.D.)--North Carolina State University, 2019. |
| 일반주기 |
Source: Dissertations Abstracts International, Volume: 81-05, Section: B.
Advisor: Vatsavai, Ranga |
| 이용제한사항 | This item must not be sold to any third party vendors. |
| 요약 | Software analytics is routinely used by researchers and industrial practitioners for many diverse tasks. Large organizations such as Microsoft routinely make practice data-driven policy development where organizational policies are learned from an extensive analysis of large datasets. However, despite these successes, there exist some limitations to modern software analytic tools - 1. Lack of relevant data to perform the analytics, and 2. Lack of insightful analytics. This thesis attempts to highlight and offer potential solutions to these pressing problems. A premise with most of the prior work on software data analytics is that there exists data from which we can learn models. But this premise is not always satisfied. When local data is scarce, sometimes it is possible to use data collected from other projects. Researchers achieve this using transfer learning which seeks to transfer knowledge from some source project and apply it to a target project. Much of the transfer learning methodologies achieve this by using complex dimensionality transformations. However, these methodologies were seldom generalizable and needlessly complex. To address this, this thesis offers a very simple "bellwether" transfer learner. Given N data sets, we find one dataset that which produces the best predictions on all the other projects. We call this the "bellwether" data set. We show that these can then be used for all subsequent analytics. We explore the existence of Bellwethers in a number of domains within software analytics: (a) Code smells detection |
| 일반주제명 | Computer science. |
| 언어 | 영어 |
| 바로가기 | 
: 이 자료의 원문은 한국교육학술정보원에서 제공합니다. |
