Artificial Neural Network Applications for Software Reliability Prediction

Chapter 6Prediction of Software Development Efforts

To develop reliable software within given time and budget constraints, software managers need to keep a tight vigil on software quality and reliability and dynamically manage their resources. The resource management is highly dependent on quality and reliability assessment of software under development besides understanding the quality and cost of resources available. The later part is generally understood qualitatively and in general difficult to present in good quantitative terms. However, the quality and reliability measurement is challenging task for managers when product is under development. Indicative measures of software reliability and quality help to predict resource requirements for meeting the specified performance requirements. Software development efforts prediction is one of the important activities for the above purpose. Inaccurate prediction of efforts may lead to cost overruns or poor reliability of software or both. Overestimation of the efforts leads to wastage of software development resources and underestimation of the efforts causes schedule delays, poor quality of software and associated penalties.

ANN has been used in various areas in the last two decades. It is used for predicting cumulative number of failures, time between failures and fault-prone modules in a software (Jin 2012, Jin 2011, Mohanty 2013, Changjie 2011). It has been found that ANNs have better predictive capability than most conventional ...