International Software Measurement & Analysis Conference
Presented by IFPUG

Track Presentation Abstracts

Estimating Latent Defects Using Capture-Recapture: Lessons from Biology
Joe Schofield, Sandia National Laboratories

Statistical sampling techniques for populations in biology can be easily applied to peer reviews and inspections to estimate latent defects in software products. In turn, these values can be used to quantify the quality of the process and to establish thresholds for repeating review and testing practices. Fifth graders have demonstrated competence in using CR method after a short introduction. “Participants” in this session will get hands-on experience in using CR enabling them to help target effective defect-removal processes in their organizations. This approach can be used to support measurement-related CMMI(R) ML 2, 3, and 4 practices.

Highlights:

• Recent defect horror stories
• Why testing will never eliminate defects
• How to remove defects early and reduce costs and rework
• What is Capture-Recapture Method
• Applying CRM to help your projects and organization

Call 911 – Measurement Needs CPR
Agnes E. Nanu, Booz Allen Hamilton

Extensive literature has been written about measurement and how to deal with it. And yet, more than often, organizations are still struggling with implementing a successful measurement program. Ultimately the measurement program loses its credibility and slowly but surely dies. How did this happen? What went wrong? What should we do to bring measurement back on track? What is an indicator? What is a meaningful indicator? This presentation addresses these questions and more by describing methods to approach, assess, and revitalize an existing measurement program by addressing its weaknesses.

Highlights:

• Learn how to implement the CMMI and PSM key measurement concepts
• Learn how to assess a measurement program that is already in place, and what to do to bring the measurement program back on a successful track
• Learn how to find your way in the labyrinth of measures and how to differentiate indicators from base measures and derived measures
• Learn how to identify and design meaningful measures/indicators
• Learn how to implement a meaningful measurement program

How to Estimate Software Size and Effort in Iterative Development
Dr. Ales Zivkovic, University of Maribor

Iterative development has become the predominant development approach. While providing several benefits for developers, iterative development makes size and effort estimates more difficult. The main problem is incomplete artifacts between iterations. In the proposed presentation, an approach that enables early size estimation using Unified Modeling Language (UML) artifacts will be presented.

Highlights:

• The difference between iterative and non-iterative estimation. How different abstraction levels influence estimation
• How to overcome the problem of missing data
• How to make size and effort estimates in iterative development using UML models
• Open problems and things we should be careful about

Function Points in Contracts – The Good and the Bad
Christine Green

This presentation will talk about the benefits and pitfalls of using Function Points in a contractual agreement. Is it a good idea? Without any doubt – if used correctly. The presentation will discuss “What Function Points in a contractual agreement can do for you; what you need to investigate and create before the agreement is signed; what happens with the Function Points when the agreement has been made; How Function Points can and should be linked to price.

Highlights:

• FP in Contracts
• Metrics in Contracts
• Why use Metrics in Contracts
• Business Benefits of Metrics in Contracts
• Process for Metrics in Contracts

Model for Estimating Large Scale Data Conversion Efforts
Sagar Gollapudi, Satyam Computer Services

Estimating for large data conversion projects is complex and as difficult as any other data related project. While the base conversion can be suitably estimated, the extent and cost of testing is complex and can change the entire estimate. This presentation will reflect on the work experience gained during a large data conversion effort and present the derived empirical estimation model.

Highlights:

• Parameters for sizing data conversion efforts
• Parameters for determining complexity
• Effort Estimation guidelines
• Identifying work suitable for offshore development
• Developing globally leveraged cost estimates

Using Metrics to Develop a Software Project Strategy
Donald Beckett, Quantitative Software Management

Virtually all organizations involved in manufacturing, sales, or service gather data on their activities for performance analysis, to determine their competitive position, and identify strengths and weaknesses. In software development this is often done haphazardly – if at all. As a result, misconceptions abound about the relationships between cost, schedule, and quality that result in the same mistakes in planning and execution being repeated over and over again. Using Metrics to Develop a Software Project Strategy identifies and places boundaries around the trade-offs between cost, schedule, and quality that conference attendees will find useful in their own work environments.

Highlights:

• State of the Industry: Project Estimation Modeling projects based on projected size growth
• Staffing and effort: the dangers of industry averages
• The non-linear schedule/effort tradeoff
• Why Analysis effort counts: a case study

Goal Driven Performance Measurement
Terry Vogt, Booz Allen Hamilton

Performance measurement often begins without clarity on the reasons for measuring, which can lead to misdirected effort, measuring too many variables, and measures that neither inform nor guide actions to improve. Equally problematic is the misconception that performance measurement itself will solve problems without a plan for analysis and effective action in response. Lack of effective involvement in the requirements, design, process and presentation of measurement information by key shareholders compounds these problems. This presentation will illustrate an approach to avoid these pitfalls and ensure development of effective measurement programs based on specific goals and objectives.

Highlights:

• Fundamental problems of measurement design
• How goals determine measurement Requirements
• The need to prioritize objectives
• The Goal-Question-Indicator-Measure Framework
• The Measurement Design Process

Monte Carlo, Resampling, and Other Estimation Tricks
Mauricio Aguiar, TI Metricas

This presentation will introduce participants to the use of Monte Carlo and other related techniques in software estimation. You will learn how to solve potentially complicated estimation problems using only simple procedures and Microsoft Excel.

While common statistical techniques require strict assumptions such as normality, software development data will rarely satisfy those assumptions. Techniques such as Monte Carlo, Resampling, and the Jackknife will produce good results even in “non-normality” situations. Learn how to obtain statistically valid results without tedious formulas and tables.

This presentation builds on a previously successful IFPUG ISMA presentation (Las Vegas, 2007).

Highlights:

• How to account for uncertainty in estimation inputs
• How to drop unrealistic assumptions and still be safe
• How to obtain valid statistical results without formulas and tables
• How to tackle complex problems using simple techniques
• How to do Monte Carlo with MS-Excel, VBA, and low-cost add-ins

Measurement Reports
Erika Vintan, BMO IT Best Practices & QA

This presentation describes how the measurement process was rolled out in a financial institution that became a strong CMMI Level 4 organization by using Measurement and Analysis to be proactive in influencing decisions. The presentation will focus specifically on sharing best practices and lessons learned in creating measurement reports while offering a view of the challenges faced and the challenges that still lie ahead.

Highlights:

• Why to measure
• What to measure
• How to collect data
• How to report data
• How to turn Metrics into Assets

Enterprise IT Measurement
Peter Thomas, IBM

The use of project teams located at multiple locations requires a sophisticated IT Measurement data capture and repository infrastructure. Effective management requires a sophisticated data analysis and reporting capability. This presentation shares the lessons learnt by part of IBM in implementing this sophistication.

Highlights:

• The components of an Enterprise IT Measurement system
• Pros and cons of different infrastructures
• Hints and tips with identifying measurement data and dimensions
• Hints and tips with data analysis
• Hints and tips with reporting

Implementing Agile Measurement Program
Jaisankar Muthukrishnan, Accenture

Software organizations often are challenged with getting a right measurement program for the various types of work. Often various stakeholders perceive that projects collect lots of metrics but the metric program defined is complex, difficult to understand and see no value in it. This presentation will address how an agile measurement program can come to an organizations rescue in addressing the challenges outlined above. We will discuss what an agile measurement program is, essential principles, techniques to ensure data quality, roles of users and SEPG teams and implementation guidelines for the program.

Highlights:

• Understand the expectations of the various stakeholder groups in the measurement program
• Gain insights on implementing comprehensive measurement programs addressing various types of work
• Understand the tactical tool implementation techniques
• Methods to drive away data quality issues
• Building steps for comprehensive metrics analysis

FPA Applied with UML/Use Case
Adri Timp, Equens

FPA is designed to be independent of technology and analysis/design-methodologies. However, for some people it is a challenge to apply FPA in new environments. They could argue that FPA is not applicable to UML/Use Case documented systems, or even promote Use Case Points instead. For an experienced counter there is no real difference between counting functions in a classical environment and in a UML/Use Case environment. The major difference is the terminology and the structure of the documentation.

This presentation provides a translation mechanism for concepts and terminology used in UML (Unified Modeling Language) designed and documented information systems into the concepts of FPA.

At first the presentation explains which UML products may be useful to apply FPA. After that it focuses into the Use Case Model and shows how to identify the transactional functions and how to apply the transactional function rules. Then it explains the Class Model and shows how to identify the data functions and how to apply the data function rules.

This study has been carried out by a NESMA task force. Mid 2008 it will be published in a handbook. The approach is universally applicable, so can be used in conjunction with the IFPUG Counting Practices Manual.

Highlights:

• UML - User functionality in new bags
• Apply FPA using UML documentation
• Translate UML concepts into FPA language
• Count transactional functions using a Use Case Model
• Count data functions using a Class Model

Relationships Among Software Metrics in Benchmarking
Scott Goldfarb, Q/P Management Group, Inc.

Determining the relationships in software measurement data requires more then statistical analysis. It requires an understanding of software development, interaction of variables and the historical trends in the software industry. With this knowledge, functional size can be correlated to productivity, cost, schedule and staffing mix.

This session will present both the characteristics that need to be considered when analyzing software metrics and the results of statistical analysis when the appropriate analysis is performed. The characteristics and variables that will be discussed that influence productivity and cost will include outsourcing, offshore development, schedule duration, platform, function point size, staffing levels, process and quality.

Highlights:

• What attributes need to be considered when analyzing measurement data
• How industry trends have influenced performance over the years
• The major factors that impact productivity and cost
• How to organize project data to determine relationships
• How to utilize benchmark data

Using Function Points for Project Tracking and Earned Value Analysis
Steve Keim, The David Consulting Group

The basis for Earned Value Analysis is that the work that you have performed up to now has a measurable dollar value to the entire project. Function points are an effective way of establishing and tracking that value.

Highlights:

• Function Points list the business functions for your application in business language.
• Function points are applicable to all phases of the development life cycle.
• You can easily identify and track progress for each business function. How to allocate value within your organization.
• How to compute Earned Value using a phased functional approach.

Can SAP be Function Point Counted?
Debra Maschino, Q/P Management Group, Inc.

Have you even wondered how to count SAP? Where do you start? How do I identify the files and functions? This presentation will address all these questions and more related to the counting of COTS products and other ERP systems with their parameter driven functionality and flexibility.

Highlights:

• Where to start?
• What to count?
• How to proceed?
• Who is needed to count?
• Why would I want the size?

How to Avoid Traps in Contracts for Software Factory Based on FP Metric
Claudia Hazan, SERPRO

This work is to clarify questions on the use of the FP metric of contracts in software and on some concepts inherent in the counting of function points, such as elementary process, external output (EO) external inquiry (EQ), data conversion functions, general system characteristics (GSCs). In addition, this presentation seeks to present guidelines and tips to support enterprises in the preparation of their contracts for software, aimed at reducing the conflict between contractors and clients; specifically, considering issues such as metrics for maintenance of software, metrics to changing requirements, productivity of the software development process, estimates and feasibility study for software projects.

Highlights:

• Function Point Software Contracts
• Software Maintenance Metrics
• Requirements Change Metrics

Are You Managing IT, or is IT Managing You?
David Garmus, The David Consulting Group

In 1994, the Standish Group published its first CHAOS Report on the state of software development. Their most recent report in 2006 indicated only modest improvement: only 35% of software projects were successful; 19% were total failures; and 46% were challenged by being over calendar time and cost and under quality expectations. How is your organization doing in managing your IT projects? David Garmus will provide some milestones for you to evaluate your success against the success of others in the industry.

Highlights:

• Measurement must be matched to business user needs
• A continuous, methodical process for comparing performance levels among projects, divisions or organizations is Essential
• Identification of critical quantitative & qualitative factors
• Mapping attributes based on organization’s tolerance for risk and change
• Industry benchmark data will be provided to support project productivity measurement

Counting Web 1.0, 2.0, 3.0
Raymond Boehm, Software Composition Technologies

Since the 80s, the web has been evolving in terms of its capabilities and use. Web 1.0, 2.0 and 3.0 refer to different major generations of web applications. This talk will describe the differences between these generations. There are different principles involved in counting Web 1.0 and 2.0 applications. These differences will be discussed, along with observations regarding some counts that have been performed. Web 3.0 is too new to have generally accepted counting principles. However, some speculation regarding principles for this emerging generation will be presented.

Highlights:

• Definitions of Web 1.0, 2.0 and 3.0
• Retrieval (Web 1.0) FP principles
• Interactive (Web 2.0)
• FP principles Representative counts and significance
• What to measure in Web 3.0 sites

Measuring the Business Value of IT
Michael Harris, David Consulting Group

How should we run our IT Departments to ensure value for money? It’s a question that is on everyone’s mind, from board rooms to business managers. Interestingly enough, the question itself contains the key phrase that unlocks the mystery – if we can identify ways to measure value and prioritize the metrics identified then we can build processes that generate the desired behavior and metrics.

All services are measurable and when we view IT as a service provider we can begin to define the boundaries of how IT can be measured.

The intent of this presentation is to provide answers to questions that we have commonly heard from CEO’s, CFO’s, CIO’s down to development directors and individual project managers.

Highlights:

• How do I measure the value of IT?
• What should the business expect from IT?
• What frameworks should IT use?
• How do I measure IT Performance?
• What should IT expect from the business?

Counting Infrastructures
Dr. Anthony L Rollo, CIT Ltd

Many function point counters and managers of software counts believe that only whole applications may be sized using the IFPUG method. This is sometimes cited by some other methods as a reason for not using IFPUG. I and some colleagues have however been required by our clients to arrive at sizes for major software components such as the HCI component, the Server component and other software items often regarded as infrastructure.

This talk will outline the method of undertaking such counts. It will also deal with the important issues of identifying users, boundaries and so on from counting such software. The talk will also deal with the reasons why such counts are useful to the software developers. There can be pitfalls when reporting such counts to clients and counters need to be aware of these so they may avoid them.

Highlights:

• That the IFPUG method is more widely applicable
• The essential issues when counting components
• Why count components Why count infrastructure
• Possible pitfalls when reporting such counts

Hammer’s PEMM for CMMI
Rob Donnellan, Q/P Management Group, Inc.

Today, many executives accept as common knowledge that process-based transformation can result in significant performance improvement. But in 1990 when Dr. Michael Hammer described process management in his article “Reengineering Work: Don’t Automate, Obliterate;” the concept was considered controversial.

The Process and Enterprise Maturity Model (PEMM) examines two categories of characteristics to determine the barriers to change, what needs to change, how much change is needed, and when. One category applies at the process level, and contains the enablers (such as metrics, owners, and performers) needed for process health. The second category includes the organizational capabilities (such as culture and governance) required to support healthy processes.

This presentation will describe the PEMM framework, discuss its application to CMMI process transformations, describe companies’ experiences with it, and point to where to find more information about it.

Highlights:

• Background for the development of the PEMM framework
• A high level understanding of the PEMM framework
• How to apply the PEMM framework to CMMI process implementations
• How companies have used the PEMM framework
• Where to find more information about the PEMM

FP Sizing: Hard Lessons Learned
Swaminathan Subraman; Ajay Verma; Faustin Moses Rethinam

Counting complex applications has always been a challenge for FP counters. With the increasing demands for quicker turnaround of projects and applications from customers and pressure of improving productivity, the function point counters are expected to be consistent and fast in their counting decisions, all the while adhering to the CPM guidelines. This presentation is an attempt to explain the various scenarios encountered while sizing the application.

Though IFPUG has released various versions of CPM since inception, still counters face tricky situations in the application of IFPUG rules on current Emerging Technologies.

We will discuss the following situations and Emerging Technologies and suggested FP Counting solutions:

• Reusable components
• Packaged Applications
• Files or tables maintained in the data repositories of packaged applications
• Interfacing of applications with Legacy Systems
• Enhancements of packaged applications

A Case Study of a Successful Measurement Program as a Key Input Into Improving the Development Process
Pam Morris, Total Metrics

The presentation reviews the implementation of measurement within a large government department and how the measures have provided new insights into their development and enhancement processes.

The original object of the measurement and benchmarking activity was to assist the IT Department in demonstrating the cost efficiencies and benefits achieved by the gradual re-factoring their large legacy application (14,000 fps) over a period of 4 years.

Although the demonstration of the efficiencies of the re-factored functionality was the initial primary objective of implementing measurement, the process of baselining, collecting and analysing the measures has provided insights into the Departments development processes that have enabled significant productivity and quality improvements within their current system. These insights have changed the way that the Project managers plan and estimate their ongoing change requests that make up their quarterly Release packages. They have made measurement one of the major inputs into their decisions on where to focus their improvements.

The presentation looks at what are the key success factors of this measurement program that has been now going successfully for over 2 years and how senior management have used the data in their decision making.

Counting Facebook
Thomas Cagley, David Consulting Group

Counting Facebook will challenge counters of all levels by confronting them with scenarios that are found in the Facebook environment. Tom will present situations from normal interactive web pages to mashups of applications that will help counters interpret the CPM as technology evolves how we present information and interact.

Highlights:

• How to count embedded applets.
• Effect of mashups on boundaries.
• An approach to count user introduced changes.
• Difference between delivery and productivity.
• Recognizing transactions and files in WEB 2.0 applications.

The Myths, The Methods, The Magic
Lori Holmes, Q/P Management Group, Inc.

Baseline studies are a useful tool for organizations. They can provide information for Outsourcing, Process Improvement, and Portfolio Management, plus more. This presentation will dispel the myths related to baseline studies (e.g. they take too long); discuss methods for conducting a baseline study; and provide the magic (e.g. what can you do with the results)

Highlights:

• How to define the purpose of the baseline study
• How to determine the appropriate Sample size
• How do define and collect the data
• Examples of analysis and reporting
• Example of how to use the information

Guerrilla Metrics
Bill Hufschmidt, Development Support Center, Inc

• Inexpensive, overlooked Savings and Value Metrics that will make “them” sit up, take notice and save your job.
• Vendor Comparisons: Vendor A charges $75/Hr. Vendor B charges $50/Hr. Which is the better Value? Prove it!
• Cost of Quality: How much does a defect really cost? How can I prove a $1M savings?
• Churn: What is it? How do I measure it? Can it prove a $25M savings in Operations?
• Unreported Savings:
• If sales increased by 24%, would your CEO be happy, impressed?
• If your organization produced 24% more product at no extra cost, would your CEO be happy, impressed?
• How can you be the messenger of such savings in your organization?
• Lessons Learned

Back to Top