Estimation maturity model using function points
-
Upload
bhupinder-singh -
Category
Software
-
view
49 -
download
0
Transcript of Estimation maturity model using function points
Improving Estimation Organizational Maturity
Achieving project success and productivity improvements with mature estimation practices
1
Contents
2
Estimation Organizational Maturity Model (EOMM) What is the EOMM?
The 5 stages of Estimation Maturity
Key differences between Level 1 and Level 2
Achieving Level 2 maturity using Function Points Introducing Function Points
Function Points across the project lifecycle
Associated Metrics
Resources
What is the EOMM?
3
Proposed by Dan Galorath, chief architect and developer of SEER, software cost estimation tool Defines 5 levels of estimation maturity
Most organizations are at Level 1
Provides a roadmap to Improve estimation accuracy
Improve project predictability
Measure and improve productivity
Level 0Informal
Manual Estimation
with no process
Level 1Direct Task
Estimation (Work Breakdown Structure)
Spreadsheets Ad Hoc Process
Level 2Formal Sizing (e.g. Function
Points)
Direct Task Estimation
Simple Model (Size*Productivit
y)
Some Measurement and
Analysis)
Informal Process
Level 3 Formal SizingRobust
parametric Estimation
Estimation vs Actual
capture
Formalized multiple estimate process
Rigorous measurement and Analysis
Risk Management
Repeatable Process
Level 4 Formal Sizing Repeatable Process
Robust Parametric Estimating
Rigorous Measurement and Analysis
Parametric Estimation
with Tracking and control
Risk Management
Process Improvement
via lessons learned
Level 5 Formal Sizing Repeatable Process
Robust Parametric Estimating
Rigorous Measurement and Analysis
Parametric Estimation
with Tracking and control
Risk Management
Continuous Process
Improvement
The 5 Stages of Estimation Maturity
4
Source: Galorath.com
Key Differences between Levels 1, 2 and 3
5
Level 1 is based on Work Breakdown Method
Level 2 is based on formal software sizing Size can be expressed in units like Lines of Code (LOC), feature points,
use case count or Function Points
Introduces basis for organizational productivity measurement for application development
Level 3 lays the basis for continuous process improvement Uses parametric estimation (like COCOMO)
Process improvement leads to productivity gains
According to Gartner, just moving up from Level 1 to 2 reduces estimated vs actual variance by 50%
Difference Between Formal Sizing and WBS
6
Formal Sizing: Key Features Effort = Size / Productivity
Size is measured from a user’s perspective
Productivity depends on technology/skill level/ organization environment and other factors
A large number of secondary metrics can be predicted
Large industry database available to benchmark
WBS: Key Features Directly provides estimates in person hours (or days/months)
Based on developer’s perspective, varies by person
Productivity is hidden and cannot be verified, measured or improved
Does not factor in organizational productivity
Parametric Estimation
7
Level 3 introduces parametric based estimation Factors in a number of attributes that determine productivity
COCOMO is an example of a parametric based estimation technique
Since productivity is broken down into smaller attributes, it is easier to improve productivity by managing those attributes
How to measure software size
8
Number of sizing options available Function Points
Others like SLOC, feature points, use case count
Function Points Industry Standard, developed by IBM in 1979
Rich industry database base
Can be used to compare projects and applications
Rich knowledge base of secondary metrics available
Moving to Level 2…introducing Function Points
9
Using Function Point (FP) sizing at project start and during execution can assist planning accuracy, manage and improve project delivery
This applies only to projects where FP sizing can be used (Application Development and Enhancements, no infrastructure projects)
Using Function Points across the project lifecycle
10
Phase Key Derived Metrics Common across phases
Scoping High Level Effort, Cost and Duration, Risk Assessment
• Productivity• Drive Standardization
Requirements Baseline Effort, Cost and Duration
Design and Development Scope Creep
Testing Defect Count and Rework Effort
Maintenance Defect Prediction, Cost of Maintenance
• Cost• Duration• Scope Creep• Risk Assessment• Defect Prediction• Rework Effort
Size Application(FP)
Historical Data
Scoping
11
Take Decisions Early WBS based estimates take long to compute as they need inputs from many teams
FP Sizing is a lot quicker and helps negotiate scope w.r.t duration, scope and cost
Estimating with limited information Quick FP and assumptions can help in determining initial FP count
Historical data for duration and cost can be leveraged
Determine size by functional modules and provide an overall cost map (size is a good measure of cost)
Other benefits Assess Replacement Impact Compare FP size of existing and replacement systems
Compare Vendor Solutions
Requirements
12
Determine baseline Effort, Duration, Cost
Phase wise planning using historical data
Evaluate Requirements Completion (e.g. use industry or own historical data)
Determine Risk (e.g. Projects less than 500 function points have a risk of failure of less than 20% while those over 5000 function points which have a probability of cancellation close to 40%.)
Quantify scope for phased development using quantified ‘what if ’ scenarios
Development and Testing
13
Determine FP count at phase end to Measure scope creep
Compare with baseline FP count at end of requirements phase
Predict Defect Count at each phase
Defect prediction can be used to assess project risk to meet expectations
Associated Metrics: Effort and Duration
14
Barry Boehm’s formula to compute duration as a function of effort
Calendar Months (A) = 2.5 x Person Months 1/3
Customize this equation based on historical data:– Calendar Months = A + coefficient
Defect and Productivity Related Techniques
15
There are techniques that can be used (and improvised) for defect prediction when sufficient historical data is not available Count of UAT test cases = 1.2 x FP
Total count of test cases = FP1.2
Productivity COCOMO can be used to fine tune productivity numbers
Source
Resources
16
Galorath Inc
SPR Estimation Resources
Uses and Benefits of Function Points
COCOMO
Defect Prediction and Prevention
17
Thank you
Bhupinder Singh
http://ca.linkedin.com/in/bhupindersingh