The continuing poor performance results produced by statistical cost estimation models have flooded the cost estimation area for over the last decade. Their inability to handle categorical data, cope with missing data points, spread of data points and most importantly lack of reasoning capabilities has triggered an increase in the number of studies using non-traditional methods, the latest of which is machine learning techniques. This technique categorises together the methods:
Machine learning is a new area which is demonstrating the promise of producing consistently accurate estimates [WITTIG et al.]. The system effectively "learns" how to estimate from training set of completed projects. In theory this approach is more robust against noise and data outliers all suggesting machine learning techniques are suitable for software project effort prediction. This is exemplified by a research project undertaken at Bournemouth University by Schofield involving the development of an analogy based case tool called ANGEL. This tool has produced accurate results and a cut down version is available
Click on the link below to see a comparison of each of these machine learning techniques.
Cost Estimation Sites on WWW
Software Metrics Research Lab
http://divcom.otago.ac.nz:800/COM/INFOSCI/SMRL/home.htmThe SMRL carries out research on software metrics, data analysis and modelling techniques
(neural nets, fuzzy logic, statistical methods, case-based reasoning and hybrid-system)
