THE METHOD OF PREDICTION OF SOFTWARE PROJECTS SUCCESS BASED ON THE ANALYSIS OF PREDICTED VALUES OF SOFTWARE CHARACTERISTICS

Abstract

The task of development of methodological support for predicting of the software projects success is solved. The object of research is theprocess of predicting of software projects success at the early lifecycle stages. The subject of the research is a method of prediction of softwareprojects success based on the analysis of predicted values of software characteristics. The purpose of the work is to increase the accuracy ofselection of software projects and to increase the success of software projects. The intelligent method of prediction of software projects success is proposed. It provides to calculate the integrative indicators of success of various software projects based on the predicted (by artificial neural network) values of the characteristics of the software developed for the project. It also provides to calculate the probability of success ofsoftware projects. The method differs from the known methods, that provides to predict the success of software projects, compare softwareprojects on the main characteristics of the project and the software and on the predicted value of the probability of success (and not just onthe cost and duration, as is happening now), and perform the grounded choice of software project by customer and by developer for furtherrealization. Experiments on investigations of the method operability were performed The experimental results provide to recommend theproposed method for use in practice.