Article Preview
TopVarious approaches have been adopted in literature to make an accurate estimation of software effort. In recent years, application of machine learning approaches has been attempted. This has been possible due to the availability of data sets of a large number of completed projects. Among various machine learning techniques, neural networks based models are newly emerging models. Though several researchers (Venkatachalam, 1993; Finnie et al., 1997; Samson et al, 1997; Lee et al.,1998; Heiat, 2002; Ideri et al, 2002; Idri et al, 2004; Tadayon, 2005; Idri et al, 2006; Kanmani et al, 2007; Tronto et al., 2007; Park and Beak, 2008; Tronto et al, 2008; Iwata et al, 2009; Reddy and Raju, 2009; Ajitha et. al, 2010; Kaur et al, 2010; Bhatnagar et al, 2010; Balich and Martin, 2010; Reddy et al, 2010; Pendharkar, 2010; Attarzadeh and Ow, 2010; Dave and Dutta, 2011a, 2011b, 2011c; Shepperd and MacDonell, 2012; López-Martín, 2014) have worked on Neural Networks based models. However, there is still a need for more research work and attempts to find out the most suitable model for software effort prediction in term of accuracy, configurability and reasoning associated with the suitability of model. The detailed review of 21 articles related to the effort/cost estimation of software using neural networks is presented in Dave and Dutta (2014). The paper also suggests to research on neural networks based models for more accuracy.