Survival analysis for the duration of software projects

Abstract

In the area of software engineering various methods have been proposed in order to predict the cost of a software project in terms of the effort or of the productivity. An important feature which is closely related to the cost is the duration of a software project. In this paper we deal with the problem of studying and modeling the distribution of the time from specification until delivery of a software product. Specifically, we investigate the use of a statistical methodology known from biostatistics as survival analysis. The purpose of such an analysis is to describe the distribution of the duration and also to identify important factors that affect it. The great advantage of survival analysis is that we can utilize information not only from the completed projects in a dataset but also from ongoing projects. The general principles of the methodology are described with examples from applications to known data sets