ABSTRACT
The main purpose of this chapter is to demonstrate that machine learning-based software project cost estimation models built on the basis of software Functional Size Measurement (FSM) methods approved by ISO/IEC and using benchmarking data of sufficient quantity and quality can help to reduce the practical problems arising from incorrect cost estimation of software projects. This is significant to limit the losses of often exceptionally large funds invested in IT/software projects, especially in the case of large projects financed from the state budget, where the waste of funds for individual projects can amount to even several hundred million dollars. The chapter presents both the conditions necessary to build such effective models as well as the conclusions resulting from their use, which can provide a decision-making support for organizations developing and implementing software systems.
