| Title | Developing Statistical Software in Fortran 95 |
| Author(s) | David R. Lemmon, Joseph L. Schafer |
| Publisher | Springer |
| Year | 2005 |
| Edition | 1st edition |
| Language | English |
| Pages | 323 |
| ISBN | 9780387281230 |
| Series | Statistics and Computing |
| Genre | Statistics, Computer Science, Programming |
| Size | 5.97 MB |
| Extension |
Summary
Developing Statistical Software in Fortran 95 is a specialized guide that fills a significant gap in the literature on computational statistics. Published by Springer in 2005 as part of the Statistics and Computing series, this book was written by David R. Lemmon and Joseph L. Schafer, two researchers affiliated with The Methodology Center at The Pennsylvania State University [citation:1][citation:3]. While many texts teach the theory of computational statistics, very few address the practical challenge of writing well-engineered, maintainable programs. This volume directly tackles that challenge, offering statisticians, biostatisticians, and methodologically oriented researchers a structured path to developing high-quality statistical software using modern Fortran [citation:1][citation:3]. The authors bring complementary expertise to the project: Lemmon’s background in engineering and statistical programming, and Schafer’s distinguished career in statistics and missing-data methodology, ensuring that the content is both technically sound and practically relevant.
The book is organized around a progressive, hands-on approach to software development. It begins with an introduction to Fortran 95 and advocates a pseudo object-oriented programming style, allowing readers to structure their code in a modular and reusable fashion even within the constraints of the language [citation:3]. From there, it moves into the implementation of computational routines, emphasizing accuracy and efficiency in numerical procedures. A substantial portion of the text is devoted to the practical deployment of these routines: creating console applications, building dynamic-link libraries (DLLs), and developing COM servers and clients [citation:3]. This emphasis on integration and deployment distinguishes the book from purely theoretical treatments of Fortran programming, as it prepares readers to make their computational work accessible to a broader audience.
One of the most valuable aspects of the book is its extensive use of detailed examples that demonstrate how to call Fortran procedures from widely used software packages, including Excel, SAS, SPSS, S-PLUS, R, and MATLAB [citation:1][citation:3]. These examples bridge the gap between the Fortran environment and the tools that most statisticians and researchers use in their daily work. The book even includes a tutorial on creating graphical user interfaces for Fortran computational code using Visual Basic.NET, further extending the reach of the developed software [citation:3]. For readers who already have some familiarity with programming, the book offers strategies for harnessing the power of modern Fortran and the object-oriented paradigm, while those with less experience will find that prior knowledge of a language such as Basic, Fortran, or C is helpful but not strictly required [citation:1].
The target audience for this volume is clearly defined: statisticians, biostatisticians, and methodologically focused researchers who need to create statistical applications quickly and effectively. It is also well suited for use as a supplementary text in graduate courses on statistical computing [citation:1][citation:3]. The book assumes a basic level of quantitative and programming literacy but does not demand extensive prior experience, making it accessible to graduate students and researchers who are new to software development in a statistical context. More seasoned programmers will still find value in the book’s treatment of modern Fortran features, COM interfaces, and the practical architecture of statistical applications, areas where the authors’ combined experience offers insights that are difficult to find elsewhere [citation:3].
Since its publication, Developing Statistical Software in Fortran 95 has been recognized for its unique contribution to the field. Reviews in journals such as Technometrics, the Journal of the American Statistical Association, and the Journal of Statistical Software have praised its practical focus and clear organization [citation:3][citation:8]. One reviewer described it as “the best introduction to programming that I have ever read,” while another noted that the book’s real strength lies in its guidance on building COM interfaces, making it especially useful for those who need to create such interfaces [citation:3]. The book remains a valuable reference for anyone interested in the intersection of statistical methodology and software engineering, offering a perspective that is grounded in real-world application rather than abstract theory. Its enduring relevance stems from its focus on principles of good programming practice and integration, which transcend specific language versions and remain applicable to modern statistical computing challenges.
Key Features
- The book explicitly teaches a pseudo object-oriented programming style for Fortran 95, enabling readers to write modular and reusable statistical code.
- It provides a complete roadmap for developing statistical software, from writing computational routines to building console applications and DLLs.
- Detailed examples show how to call Fortran procedures from Excel, SAS, SPSS, S-PLUS, R, and MATLAB, making statistical routines accessible to users of these platforms.
- A dedicated tutorial guides readers through creating graphical user interfaces for Fortran code using Visual Basic.NET.
- The text includes coverage of COM servers and clients, allowing Fortran computational components to be integrated into Windows-based software architectures.
- It is written specifically for statisticians and biostatisticians, with examples drawn from statistical computing rather than generic programming exercises.
- The book assumes no prior Fortran 95 experience, making it suitable for researchers who are proficient in other languages but new to modern Fortran.
- Each chapter includes exercises that reinforce the concepts and techniques presented, supporting both self-study and classroom use.
- Code examples from the book are available for download, allowing readers to compile and experiment with the programs on their own systems [citation:15].
- The authors bring real-world experience from The Methodology Center at Penn State, where they developed statistical software for research and education [citation:1].
- The book has been reviewed favorably in major statistical journals, including Technometrics and the Journal of the American Statistical Association [citation:3].
- It addresses the complete software development lifecycle for statistical tools, from algorithm implementation to user interface design and cross-platform integration.
About Author
David R. Lemmon is a Research Associate at The Methodology Center in The Pennsylvania State University’s College of Health and Human Development. He holds a Ph.D. in Mechanical Engineering from the University of Cincinnati, and his earlier research focused on finite element analysis, automobile crash simulation, and biomechanical modeling of the diabetic foot [citation:1][citation:9]. Since 2001, he has served as the Center’s Senior Statistical Programmer, where he manages the creation of new statistical software for research and education [citation:1]. His transition from engineering to statistical programming gives him a unique perspective on software development that bridges numerical computation and applied statistics. His work at The Methodology Center has involved developing tools for the analysis of missing data and other methodological challenges, areas in which he has collaborated closely with Joseph L. Schafer.
Joseph L. Schafer is an Associate Professor in the Department of Statistics at The Pennsylvania State University and a Principal Investigator at The Methodology Center [citation:1]. He earned his Ph.D. in Statistics from Harvard University in 1992, where his doctoral advisor was Donald B. Rubin, a leading figure in missing-data methodology [citation:10]. Schafer’s research focuses on statistical methods for missing values, the analysis of longitudinal data, statistical computing, and methodology for studies of substance-use prevention [citation:1]. He is widely known for his work on multiple imputation and has developed techniques for analyzing incomplete data and incorporating missing-data uncertainty into statistical inference. He has served as a statistical consultant to numerous government agencies and has taught workshops and short courses on missing data topics. In 2019, he was named a Fellow of the American Statistical Association, recognizing his contributions to the field [citation:10]. His collaboration with Lemmon on this book reflects a shared commitment to advancing the practice of statistical computing through well-designed software.
Related Books
- Modern Fortran Explained — Michael Metcalf, John Reid, and Malcolm Cohen
- Fortran 95/2003 Explained — Michael Metcalf, John Reid, and Malcolm Cohen
- Numerical Recipes in Fortran 90: The Art of Parallel Scientific Computing — William H. Press, Saul A. Teukolsky, William T. Vetterling, and Brian P. Flannery
- Statistical Computing with R — Maria L. Rizzo
- Software for Data Analysis: Programming with R — John M. Chambers
- Computational Statistics — Geof H. Givens and Jennifer A. Hoeting
- Missing Data in Longitudinal Studies: Strategies for Bayesian Modeling and Sensitivity Analysis — Michael J. Daniels and Joseph W. Hogan
- Analysis of Incomplete Multivariate Data — J. L. Schafer
Ads
FAQ
Q : Who are the actual authors of Developing Statistical Software in Fortran 95?
R : The book was written by David R. Lemmon and Joseph L. Schafer, both affiliated with The Methodology Center at The Pennsylvania State University [citation:1][citation:3]. The old article incorrectly attributed the book to the IEEE Computer Society’s Software Engineering Standards Committee, which is a standards body and not the author of this volume. Lemmon is a Senior Statistical Programmer with a Ph.D. in Mechanical Engineering, and Schafer is an Associate Professor of Statistics with a Ph.D. from Harvard. Their combined expertise in statistical computing and software development is what makes the book’s practical guidance so authoritative.
Q : What distinguishes this book from other Fortran programming manuals?
R : Most Fortran books focus on language syntax or numerical algorithms, but this volume is specifically about developing complete statistical software applications. It goes beyond writing subroutines to address the entire development process: structuring code with a pseudo object-oriented style, creating console applications, building DLLs and COM components, and designing graphical user interfaces [citation:3]. The examples are drawn from statistical computing, and the book explicitly shows how to integrate Fortran routines with popular statistical and data-analysis packages like R, SAS, SPSS, and MATLAB. This end-to-end perspective on software development is what sets it apart from conventional Fortran references.
Q : Does the book require prior experience with Fortran or other programming languages?
R : According to the authors, prior experience with a language such as Basic, Fortran, or C is helpful but not required [citation:1]. The book introduces Fortran 95 in a way that is accessible to programmers coming from other backgrounds, and the pseudo object-oriented style is explained from first principles. That said, readers with no programming experience at all may find the pace challenging, as the book assumes a basic familiarity with concepts like variables, control flow, and subroutines. For graduate students or researchers who have used R, MATLAB, or similar tools, the material should be approachable.
Q : How does the book help readers integrate Fortran code with software like R or Excel?
R : The book devotes substantial attention to the practical mechanisms for calling Fortran procedures from other environments. Through detailed examples, it shows how to create DLLs that can be loaded by Excel, SAS, SPSS, S-PLUS, R, and MATLAB [citation:3]. It also covers COM servers and clients, which allow Fortran computational components to be used within Windows applications and frameworks like Visual Basic.NET. This focus on interoperability means that readers can leverage Fortran’s computational speed while still providing access to their algorithms through the tools that their colleagues and clients already use.
Q : What is the significance of the “pseudo object-oriented” approach mentioned in the book?
R : Fortran 95 does not have native support for object-oriented programming features like classes and inheritance, but it does provide modules and derived types that can be used to emulate some object-oriented principles. The authors advocate a “pseudo object-oriented” style that organizes code around data structures and associated procedures, promoting encapsulation and reusability without requiring features that Fortran 95 lacks [citation:3]. This approach helps statisticians write code that is more modular, easier to maintain, and less prone to errors, especially as projects grow in complexity. It is a practical compromise that leverages the strengths of Fortran while adopting beneficial design patterns from other paradigms.
Q : Is this book suitable for self-study, or is it intended for classroom use?
R : The book works well in both contexts. Each chapter includes exercises, and the code examples are available for download, making it feasible for independent learners to work through the material at their own pace [citation:15]. At the same time, the structured progression from basic Fortran 95 concepts to advanced topics like COM interfaces makes it suitable as a supplementary text for a graduate course on statistical computing [citation:1]. Instructors may find that the book’s practical orientation complements more theoretical treatments of computational statistics, giving students hands-on experience with the software development process.
Q : What has been the critical reception of this book since its publication?
R : The book has been reviewed positively in several major statistical journals. In Technometrics, Melvin J. Hinich called it “the best introduction to programming that I have ever read” [citation:3]. The Journal of the American Statistical Association noted that the book’s real thrust is building COM interfaces and that it will be most useful to anyone who needs to create such interfaces [citation:3]. The Journal of Statistical Software described it as a good resource for someone who already knows some dialect of Fortran and wants to learn about Fortran 95, though it noted that another reference manual would be needed for comprehensive software development [citation:15]. Overall, the reception has been favorable, particularly for its unique focus on integrating Fortran with other software environments.
Q : Does the book cover Fortran 90 and later standards, or is it limited to Fortran 95?
R : The book focuses specifically on Fortran 95, which was the current standard at the time of publication in 2005. However, much of the material applies to later Fortran standards (Fortran 2003, 2008, and beyond) because the core language features and programming principles discussed—modules, derived types, array operations, and interoperability mechanisms—carry forward. The pseudo object-oriented style and the integration techniques with other software remain relevant. Readers using modern Fortran compilers will find that the examples largely work, though some minor adaptations may be needed. The book’s emphasis on design principles rather than version-specific syntax gives it a longer shelf life than a pure language reference would have.
Enregistrer un commentaire
Thanks for comment