Predictive Modeling in Software Engineering using Ensemble Learning Techniques

Project: ResearchStUp

View graph of relations

Researcher(s)

  • Jacky Wai Keung (Principal Investigator)
  • David BUDGEN (Co-Investigator)
  • Ross JEFFERY (Co-Investigator)
  • Barbara KITHENHAM (Co-Investigator)
  • Ekrem KOCAGUNELI (Co-Investigator)
  • Tim MENZIES (Co-Investigator)
  • Akito MONDEN (Co-Investigator)
  • Qingbao SONG (Co-Investigator)

Description

The software engineering research community has been focusing on the application of machine learning approaches for software effort estimation for over a decade. Despite on reported successes using different prediction models, there is no consensus on which single software effort estimation methods produce the most accurate prediction result, largely due to the variations in the characteristics of the datasets being used under different experimental settings. Until recently, our preliminary experimentations on the application of multiple methods (ensembles) combining two or more solo-methods have been successfully showing significant stability and performance improvements in the models produced. The goal of this project is to continue research on the next generation of ensemble-based models for software effort estimation, this can be accomplished by evaluating on different complex ensemble schemes and to build a comprehensive ensemble construction procedure, which leads to development of more stable and accurate predictive models, an important contribution to software engineering research. The successful outcome will also radically change the way of applying machine learning techniques on software engineering data.

Detail(s)

StatusFinished
Effective start/end date1/07/1330/11/16