Browse Books

Reviewer: Sumit Sircar

Producing reliable and accurate project cost estimates has long been a dream of those concerned with project management and software productivity measurement. Roetzheim and Beasley recount how the various methods of estimating the size of a program can be utilized in a new information technology environment that includes graphical user interface and object technologies. They also show how to incorporate risk and sensitivity analysis into the estimation process. This is not, however, a general-purpose textbook. Rather, it is a description of the approach used by the authors' company and is essentially a manual for the use of their software product, COST XPERT, which is included on a compact disc, along with two of the company's other products. Therefore, a review of the book must also include a review of the software. Overall, the book and COST XPERT fulfill their main purpose, which is to help provide accurate and consistent estimates for a wide range of projects. The authors make the case that software estimation is not just an art, but a science. They describe two general approaches for software cost estimation—metrics-based and heuristic. They then advise correlating and averaging the results of multiple estimation methods. No theoretical basis for this advice is provided, however. The book's best feature is the use of COST XPERT to illustrate solutions to a variety of practical problems. COST XPERT accommodates a wide range of input parameters and environment adjustment factors, allowing users to address their individual project environments. It is also capable of performing sensitivity analysis. In its coverage of code reuse in object-oriented development, however, the book loses its practical feel. It is, in reality, very difficult to reuse code directly from other applications. In general, the book is well organized and makes good use of examples, tables, and figures. Its division into two separate parts is useful, especially for software practitioners. The first part is a discussion of the authors' proposed software project cost estimation methods, and the second is an illustration of COST XPERT. The second part might have worked better if it had been included as the second part of the discussion rather than being relegated to an appendix. A few references seem to be missing in the discussion of Boehm's COCOMO model (for instance, in chapter 8, Table 2, on page 33, and Table 10, on page 88). The problem of software cost estimation has no neat and simple answers. Software managers, however, need an unbiased view of how their project will be traced and managed within their particular software development environment. They would like to be able to solve specific estimation problems with a specific measurement model. Generality of the model is, therefore, important. The model discussed in this book is said to achieve generality by providing an open structure in terms of its allowance for a variety of input parameters and adjustment factors. The authors help readers reflect on the practical aspects of software cost estimation, a process that may benefit both practitioners and academics. By means of articulately presented arguments, the authors offer unique insight into how to use their cost estimation model for software estimation. The book provides strong support for project management, both in principle and in practice. It is best suited for practitioners who can make use of the software.