Customer Reviews

Function Point Analysis: Measurement Practices for Successful Software Projects (Addison-Wesley Information Technology Series)

5 star:		(3)
4 star:		(1)
3 star:		(1)
2 star:		(0)
1 star:		(1)

 

 

This is an authoritative book by two authors who have a published a large number of articles on the subject.

Function point analysis is, in both my experience and opinion, one of the most accurate estimating tools a software engineer has at his or her disposal. I have previously used the constructive cost model (COCOMO) through version 2, Raleigh curves, and hybrids such as Software Estimation Analysis Tool, which computes both COCOMO and function point statistics. Granted, not all of these estimating models are equal - the Raleigh curve approach has a focus on quality and manpower loading that the others lack. But each (and many I have not cited) are estimating tools for software development.

My experience shows function point analysis to be the most accurate - if done correctly. Therein lies the crux of the matter. Function point analysis, in order to yield accurate estimates of effort, needs to be performed by certified counters and the resulting data needs to be properly interpreted. That's where this book comes in.

It starts with an overview that serves as an understandable introduction for someone who is new to software estimation, yet is interesting enough to hold the attention of more experienced professionals. The advice on preparing for IFPUG (International Function Point User Group) certification is a highlight and underscores the fact that FP counting should only be performed by certified counters.

The chapters on software measurement and executive introduction to function points are really extensions of the introduction. I found both the software measurement model in this chapter, and how to establish a world class measurement program interesting and informative.

This segues into three chapters that cover measuring and using function points, and industry benchmarks. Some highlights are the way the authors tie function points to key performance indicators such as productivity, quality, financial and maintenance. I also liked the way that the authors stratify the users of function point data into project managers, IT managers and cross-organizational stakeholders. I was particularly interested in the last group because I am often involved in both service and application delivery projects. The use of function points as a basis for service level metrics and outsourcing is an intriguing idea that I will explore further.

When you arrive at chapter 6 the foundation built in the first five chapters prepare you for a detailed, methodical approach to performing function point analysis. This is straightforward and appears to be completely consistent with IFPUG. All you need to know about function point analysis is, in my opinion, provided in the remainder of the book. The material is augmented by case studies, and also addresses contemporary issues such as applying function point analysis to object-oriented, web, client/server and data warehouse applications. While each of these topics are not given in-depth treatment the fact that they are addressed at all is impressive. The authors end the book with advice on preparing for the certified function point specialist exam, and gives sample exam questions. They also provide sample worksheets in the appendices.

This book merits a solid five stars because it is well written, up-to-date and follows IFPUG recommendations. If you are pursuing CFP certification this book will simplify your life. If you want to learn about an effective software estimating technique this book is a great starting point. To be completely fair, if you are new to software estimation you should also read Estimating Software Costs (Caspers Jones), which gives a complete and unbiased survey of all of the major (and some obscure) software estimating approaches.

This is a really useful book. It's a must if you need to look deep into the Function Point Analysis and get a clear understanding of what this is all about. As a professional and local instructor in Function Point Analysis I've got a lot of inspiration from this book for my course. I believe that beginners would gain even more from this book, however. It covers all topics from the current IFPUG Function Point Counting Practices Manual (release 4.1). It guides you all the way through the whole process and presents a lot of examples and additional explanations on the way.

Beside the operational guidelines this book also describe the Function Point Analysis in the perspective of related topics as e.g. estimation and benchmarking. Those topics are covered very thoroughly as well and the book provides a good insight in how Function Points can be used as a management tool.

In the last chapters "hot issues" as applying Function Points to GUI applications and OO-system development are discussed. Some very useful guidelines are presented here, indeed. But it's my personal experience that mapping OO components to the Function Point concept might be quite more challenging that the book indicates.

Published in 2001, this is a very detailed book on Function Points that follows the guidelines in the 4.1 version of the International Function Point Users Group (IFPUG) standard. Function Point metrics originated in IBM back in the 1970's and spread slowly out to the rest of the world in the late 70's and early 80's. As with anything originating from IBM, it's a process that's heavy on detail and requires a lot of work to get it right. Over the years, FPA has got more and more technical, to the extent that you really need to get certified in it these days (when I first took an FPA course in the early 80's, it wasn't nearly as comprehensive as it is these days, 20 years on..). The authors of this book between them have published a number of articles on the subject and have more than a few years of experience under their respective belts.

The book kicks off with an overview that provides a useful introduction for those "new" to software estimation, although it's still pretty heavy going and it assumes a process-heavy background. If you've worked for a large organization with a comprehensive PMBOK-like methodology, it'll all make sense. If not, you're going to have a headache by the time you finish the first chapter. <lol>. There's some good advice on preparing for IFPUG (International Function Point User Group) certification and the authors emphasize the fact that FP counting should only be performed by certified counters. IMPORTANT ENOUGH TO REITERATE - FP counting should only be performed by certified counters!!!! Keep this in your mind at all times. FPA is "Estimating-Heavy", not estimating-light. It don't come cheap in time or education.

The other reviewers have covered the content in enough detail that it's rather pointless to reiterate - suffice it to say I agree with their comments on the content. So, cutting to the chase, if you're looking at books on Function Point Analysis, you've obviously got a serious interest in estimating, and if so, and if FPA is where you want to go, this book is going to be very useful for you as a reference and as a guide. On the other hand, FPA is, while accurate if applied correctly (the nub of the matter.....), both time and expertise intensive - I can't emphasis the expertise factor enough. You need a sizable project with a budget sufficient to fund the additional overhead to make it worthwhile - for your average development project, this is overkill. If you've got a $50-$100 million plus project within the grasp of our sticky little paws and you can get a certified FPA resource on board, using FPA may very well help you stay within your rather sizable budget. And if your're studying towards your FPA certification, by all means delve into this book - it'll help you no end.

But having been there and done that in my past, don't try this at home kids. This book is a great resource, you'll learn all about FPA from it, but reading the book is a completely different kettle of fish to actually doing it. As another reviewer has pointed out, FPA is accurate IF DONE CORRECTLY. To get accurate estimates, you need certified FP counters, and the data needs to be interpreted correctly. Easier said than done. If you want to use FPA yourself, go study and get yourself certified before you try it for real - otherwise, you may well send your career as an estimator down the tubes.

If you've got to come up with estimates and you've got no real estimating experience, DON'T START HERE. Go look at Steve McConnell's book on "Software Estimation: Demystifying the Black Art" to get yourself started. You could also look at Caper Jones' "Estimating Software Costs" although it's a little dated and not up to date with Agile estimating practices (my bias shows through.....). (Subsequent update - apparantly -the second edition (2007) of "Estimating Software Costs" by Capers Jones addresses Agile methods...).

On the third hand, I do give this book 5 stars for what it is. Well-written, understandable, effective - and it follows IFPUG guidelines. If you're after CFP certification, this book won't lead you astray (although it is a version or 2 out of date) and it will prove a useful study aid. I've got it on my bookshelf at work and although I now use Agile Estimating Practices a la Ken Schwaber (Agile Project Management with Scrum) and Mike Cohn (Agile Estimating and Planning), I do take a look at this one every now and then just to see what the FPA guidelines are. Mind you, every time I look thru this book and think of what I might have been doing, I bless Ken Schwaber for coming up with Scrum (OK, OK, I like smaller projects and I like the Agile approach.....).

This semester I teach CSE 4322/5325 Software Project Management at the University of Texas at Arlington. I decided to understand function point analysis (FPA) much better than possible via the one- or two-slide overviews provided by Boehm, McConnell, Pressman, and other authors. I borrowed all the UTA library's books on FPA and looked to see which had information useful to me (project estimation for a computer game).

I found this book to be the most useful, mostly because it has a chapter that relates the IT-laden terminology of FPA to object-oriented terminology, which is considerably more precise. However I found the book written at a high school reading level, slow, redundant, and vague. I feel I could have distilled the data I need onto one page.

That being said, I am now fully convinced that FPA is analogous to the tale of the "Emperor's New Clothes": There is really nothing to FPA but hype. None of the books I skimmed provided any justification for why the UFP table coefficients have their values (e.g., who says a simple EO is worth 3 UFP?). I think the reason people actually go through all the trouble of learning FPA and even getting a certification in FPA (I was amazed that such a certification is actually offered) is so that they can plug into Capers Jones' data, which uses FP as one of the independent variables.

I suggest generating your own correlations (via multiple linear regression analysis) from your own organization's scatter plots of project effort and schedule versus a number of early metrics (e.g., requirements, events, states, transitions, use cases, system functions, steps in each use case). Moreover I suggest proxy-based estimation (PROBE), described in A Discipline for Software Engineering (SEI Series in Software Engineering), in which you estimate "LOC" based on a conceptual design (which is just a list of design classes, the conceptual category of each (calculation, data, I/O, logic, set-up, or text), and their estimated size class (very small, small, medium, large, very large)) PLUS your database of categorized classes from all prior projects.

Even though dated, this is still the best book on Function Points available in the English language. It complies with the 4.1 version of the International Function Point Users Group (IFPUG) standard, while the current version is 4.2. Nevertheless, the book contains many interesting hints and examples not available in the standard. You may also want to buy the IFPUG Counting Practices Manual - the official reference - directly from the International Function Point Users Group.

Sorry but now way is this book worth the price tag it carries and neither does it come close to making you any more knowleadgable than reading any of the online articles on this topic. First of all, about a 3rd of the book is nothing but how organizations can improve process using FPA measurements. Hello! if I wanted to learn about process improvements, I would look elsewhere, not in an FPA book. WHy don't you first tell us what the heck is FPA before telling us what wonderful things it can do for your organization? Finally after rambling on for 80 odd pages on process improvement we get to what exactly is FPA. And guess what, the author suddenly decides to cut down on his verbose style that he bored us with in first 80 pages. WHY? Suddenly you see copy paste of the FPA rules from the IFPUG manual with a line or two explanation of the terse rules. Explanations and examples of counting process was trivial at best and ridibulous at worst. For e.g. The author says, lets assume the report has 10 DETs and lets count 10 ! Wow, I didn't know we could count 10 like that ! Can you instead print the report and show us what fields it has, and which ILFs they came from, so we know how to count things instead of assuming counts ? And a lot of the counting chapters was just repeatitive definitions of the terse terms used in the FPA manual. Can we get explanations with examples please instead of definitions ?? I am trying to learn the concept not crack some exam.
My question to author: Why didn't you elaborate a lot more on counting process instead of wasting pages on process improvement BS? That's what the readers were paying the big bucks for.
Overall, see if you can just skim through this book at your local bookstore instead of wasting your money unless of course you have lots of money to spend!