Software Maintainability Prediction using Hybrid Data Mining Prediction Models

Software quality is a major concern in the software development lifecycle. Most software projects fail due to low software quality and inappropriate management decisions in terms of budget and time constraints. Software quality can be quantified by a number of attributes such as: usability, reliability, maintainability, etc. For instance, software defects that might lead to software failures, surely, will lead to low overall software. The focus of this proposed research is software maintainability prediction. In early research, data mining prediction models, were introduced to enable software project managers and developers to receive insights on future predicted software maintainability. Performance of these prediction models varied depending on the dataset. Hybrid prediction models were introduced in the literature of machine learning to outperform single stand-alone prediction models. In this research, we will investigate the application of hybrid models in software maintainability prediction. Software maintainability datasets will be gathered. Empirical studies will be conducted to make an empirical investigation between hybrid models and stand-alone models in terms of prediction performance of software maintainability. Results expected from these studies is to provide an empirical evidence of the superiority of hybrid models over stand-alone models in the software maintainability prediction domain.