257
www.SoftwareMetrics.Com 1 Function Point Training Instructor: David Longstreet [email protected] www.SoftwareMetrics.Com 816-739-4058

Function Point Training Instructor: David Longstreet [email protected] 816-739-4058

Embed Size (px)

Citation preview

  • Slide 1
  • Slide 2
  • www.SoftwareMetrics.Com119 Function Point Training Instructor: David Longstreet [email protected] www.SoftwareMetrics.Com 816-739-4058
  • Slide 3
  • www.SoftwareMetrics.Com120 Two Day Schedule u Day One Function Point Concepts Measurement Theory Estimating Models u Day Two Function Point Case Studies More on Estimating Models
  • Slide 4
  • www.SoftwareMetrics.Com121 Course Objectives u Understand and apply function point concepts u Understand basics of measurement theory u Introduction to software economics u Review/remember some basic statistical concepts
  • Slide 5
  • www.SoftwareMetrics.Com122 Estimate the Surface Area of a A can of diet coke
  • Slide 6
  • www.SoftwareMetrics.Com123 A different perspective
  • Slide 7
  • www.SoftwareMetrics.Com124 Who is David Longstreet u Metrics u Background u Clients u Research u Publishing u Teaching
  • Slide 8
  • www.SoftwareMetrics.Com125 My Metrics Over 2 million frequent flyer miles Consulted on every continent except Antarctica Presented papers at conferences in USA, Europe, Middle East, Asia and Africa
  • Slide 9
  • www.SoftwareMetrics.Com126 Cities
  • Slide 10
  • www.SoftwareMetrics.Com127 AndMilwaukee www.MAM.org
  • Slide 11
  • www.SoftwareMetrics.Com128
  • Slide 12
  • www.SoftwareMetrics.Com129
  • Slide 13
  • www.SoftwareMetrics.Com130
  • Slide 14
  • www.SoftwareMetrics.Com131 Website Metrics u About 15,000 unique visitors per month u Over 7,000 visitors view more than 5 pages u Over 1,000 visitors view more than 20 pages. u Free manual is viewed over 15,000 times per month.
  • Slide 15
  • www.SoftwareMetrics.Com132 Clients u Clients include Banking & Finance, Aerospace, Retailers, Animal Food, Telephony, Consulting Companies, Medical Research, Defense Contractors, Automotive, Universities, Government Agencies and others u Some clients: MasterCard, Amadeus, Ralston Purina, Lockheed, Transamerica, DirecTV, Biologic, Accenture, Motorola, Nissan, others
  • Slide 16
  • www.SoftwareMetrics.Com133 Type of Work u Productivity Assessments u Benchmark Studies u Estimating Models u Mergers and Acquisitions u Venture Capital and Initial Public Offerings u Outsourcing Agreements u Expert Testimony
  • Slide 17
  • www.SoftwareMetrics.Com134 Learning from organizations (ethology) u Collect both quantitative and qualitative data u Observe behaviors, customs, rituals, myths and ways of life u Examine artifacts and physical evidence u Build holistic picture of organization u Trend the industry as a whole
  • Slide 18
  • www.SoftwareMetrics.Com135 Worst Practices u No historical data u Failure to monitor and report status u Creating analysis documentation after coding u Excessive and irrational schedule pressures u Failure to establish clear acceptance criteria u Reduce testing time to make schedule
  • Slide 19
  • www.SoftwareMetrics.Com136 Research u Dale Jorgenson, Harvard Business School Historical Study of Productivity Rates for Software Development (1950 present). u Bureau of Economic Analysis Methods of collecting and reporting software productivity rates based upon Function Points. Measuring the IT Economy. u Securities Exchange Commission
  • Slide 20
  • www.SoftwareMetrics.Com137 Adjunct Professor (Avila University MBA & Graduate Psychology) u Industrial Organization Psychology u Managerial Economics u Statistics u Quantitative Analysis u E-Commerce
  • Slide 21
  • www.SoftwareMetrics.Com138 There are lies, dammed lies and statistics. Mark Twain Hannibal, Missouri
  • Slide 22
  • www.SoftwareMetrics.Com139 Negative things I have heard. u Software Voodoo! u Figures won't lie but liars will figure u You may prove anything with metrics.
  • Slide 23
  • www.SoftwareMetrics.Com140 Scientific Method u Knowledge comes from Systematic observation Measurement of particular variables & events u Develop both descriptive & predictive metrics. Descriptive: describes current environment Predictive: used to estimate
  • Slide 24
  • www.SoftwareMetrics.Com141 Theory of Measurement u The use of numbers to represent events, variables and characteristics. t Quantitative variables t Qualitative variables
  • Slide 25
  • www.SoftwareMetrics.Com142
  • Slide 26
  • www.SoftwareMetrics.Com143 What gets measured gets done u Partially Correct u What gets rewarded gets repeated u Measurements without consequences get ignored.
  • Slide 27
  • www.SoftwareMetrics.Com144 Background Summary u Constantly learning u Industry diverse client base u Geographically diverse client base u Research/Publishing u Growing industry knowledge u Better ways to teach and instruct
  • Slide 28
  • www.SoftwareMetrics.Com145 Measurement Theory u Introduction to measurement u Productivity u Scientific method u Historical perspective u Introduction to software economics u The idea of function points
  • Slide 29
  • www.SoftwareMetrics.Com146 Measurement Theory u Concept not new to many other disciplines Scientific Method u Relatively new to software development
  • Slide 30
  • www.SoftwareMetrics.Com147 Productivity u The output-input ratio within a time period with due consideration for quality. u Productivity = outputs/inputs
  • Slide 31
  • www.SoftwareMetrics.Com148 Improving Productivity (reducing unit costs) u Costs/FP u Hours/FP u We can mathematical reduce unit cost by t Reducing Cost t Increasing FP produced
  • Slide 32
  • www.SoftwareMetrics.Com149 Productivity u It is difficult (maybe impossible) to reduce productivity by reducing cost. t Cost / FP u In fact, for each $1 reduction in cost increases cost/fp by about $1.18
  • Slide 33
  • www.SoftwareMetrics.Com150 Effectiveness v. Efficiency u Effectiveness is the achievement of objectives u Efficiency is the achievement of the ends with least amount of resources.
  • Slide 34
  • www.SoftwareMetrics.Com151
  • Slide 35
  • www.SoftwareMetrics.Com152 International Weights & Measures u International weights and standards Standardization in 1863 u Motivation for international weights & standards was driven by trade and specifically international trade.
  • Slide 36
  • www.SoftwareMetrics.Com153 Concepts from Industrial Revolution u Management methods 1901 u Break - Even charts 1903 u Centralized accounting 1908 u Total Quality Management 1950s
  • Slide 37
  • www.SoftwareMetrics.Com154 Size of Largest Projects Since 1970
  • Slide 38
  • www.SoftwareMetrics.Com155 Hours/FP Since 1970 (Unit Cost)
  • Slide 39
  • www.SoftwareMetrics.Com156 Software Past, Present, Future
  • Slide 40
  • www.SoftwareMetrics.Com157 Failure Rates are High u 75% of metrics programs fail. u 70 80% of people who try to stop smoking do not succeed. u 60% of Alcoholics return to drinking. u 93% of of those addicted to gambling gamble again within 1 year.
  • Slide 41
  • www.SoftwareMetrics.Com158 If at first you dont succeed u The most thoroughly accepted notions in psychology is the principle that behavior eventually extinguishes if it is not followed by reward. u What gets rewarded gets done!
  • Slide 42
  • www.SoftwareMetrics.Com159 Software Economics Social Psychology Economics Organizational Behavior Statistics Psychology Software Development
  • Slide 43
  • www.SoftwareMetrics.Com160 Software Economics u Study Prices and Costs u Study Behavior u Study the whys u Comparative analysis
  • Slide 44
  • www.SoftwareMetrics.Com161 Inputs Outputs Storage All systems have..
  • Slide 45
  • www.SoftwareMetrics.Com162 Two Major Categories u Transactions Inputs Inquiries (reads) Outputs (calculations) u Storage Maintained Data Referenced Data
  • Slide 46
  • www.SoftwareMetrics.Com163 Validity of Function Points u Face Validity Does this make sense? u Predictive Validity Useful for predicting effort, time, cost, other? u Convergent Validity Do FPs move in same direction as LOC, Test Cases, Use Cases, Objects
  • Slide 47
  • www.SoftwareMetrics.Com164 Face Validity u Does this idea make sense? u Can a software application be sized by looking at transactions and files? u Are there other items we need to consider?
  • Slide 48
  • www.SoftwareMetrics.Com165 Predictive Validity u Does function points actually predict what it is suppose to predict? u To what degree (how accurate) The effort required to implement 5,000 fp's is more than 500 fps? How about 4,500 fps v. 5,000 fps?
  • Slide 49
  • www.SoftwareMetrics.Com166 Predictive Validity (other inputs) u How accurate is historical time reporting, staffing levels, defect tracking? Hours Per Function Point Duration Per Function Point Defects Per Function Point u How accurate are project plans?
  • Slide 50
  • www.SoftwareMetrics.Com167 Convergent Validity u Function Points x 1.2 approximates test cases. u Lines of Code divided by 100 approximates function points. Varies by language u Seems to be a positive relationship between test cases and function points u More staff is needed as function points increase.
  • Slide 51
  • www.SoftwareMetrics.Com168 Measurement Theory u Introduction to measurement u Productivity u Scientific method u Historical perspective u Introduction to software economics u The idea of function points
  • Slide 52
  • www.SoftwareMetrics.Com169 Understanding Software Development Costs u Large Projects u Marginal Cost u Interval Estimating v. Point Estimating u Building Estimating Models u Samples v. Populations u Industry Data
  • Slide 53
  • www.SoftwareMetrics.Com170 Large Projects u Increasing Marginal Cost As size increases unit cost rise. u Any large engineering or construction project follows the same economic model.
  • Slide 54
  • www.SoftwareMetrics.Com171 Marginal Cost u The change in total cost attributable to a one-unit change in output. u The unit cost of software is not fixed. u Unit cost changes as the size of the project changes.
  • Slide 55
  • www.SoftwareMetrics.Com172
  • Slide 56
  • www.SoftwareMetrics.Com173
  • Slide 57
  • www.SoftwareMetrics.Com174 Individuals Range of Productivity u Range of individual productivity can be 100 times t An expert developer may be 100 times more productivity than a novice. u True for other industries as well Automotive Mechanics Carpentry Cake Decorating Roofing
  • Slide 58
  • www.SoftwareMetrics.Com175 Cake Decorating?
  • Slide 59
  • www.SoftwareMetrics.Com176 Paying by the hour u What behavior is encouraged. u What type of person is hired. By the way, cake decorators charge by the piece of cake.
  • Slide 60
  • www.SoftwareMetrics.Com177 Industry Data u $1,500 function points with a margin of error of $500 t 90% confidence level u Range is $1,000 $2,000 per function u Range in hours is 100 200 hours per function point.
  • Slide 61
  • www.SoftwareMetrics.Com178 Transamerica Building u 499,000 square feet u $144 Million to construct u $290/Square Foot National Average $102/Sq. Ft San Francisco Average $140/ Sq Ft.
  • Slide 62
  • www.SoftwareMetrics.Com179 Comparative Costs Per Square Foot u Office Space $69 - $290 u Warehouse $28 - $43 u Single Unit Retail $46 - $71 u Depending on the type of software application being constructed unit costs varies.
  • Slide 63
  • www.SoftwareMetrics.Com180 Some Statistics u Range $28/Sq. Ft. - $290/Sq. Ft. u Average $75/ Sq. Ft u Standard Deviation $54/ Sq. Ft. u We see similar variations in software costs also.
  • Slide 64
  • www.SoftwareMetrics.Com181 Many other Factors u Size is size u All those other factors are part of the unit cost $/Sq. Foot $/ Function Point
  • Slide 65
  • www.SoftwareMetrics.Com182 Samples v. Populations u You do not have to FP count every single application to understand organizational productivity! u You do not have to examine every single project to understand organizational trends.
  • Slide 66
  • www.SoftwareMetrics.Com183 Gathering Historical Data Samples v. Population u Population your entire application portfolio u Sample key projects/applications to help you understand the entire application portfolio
  • Slide 67
  • www.SoftwareMetrics.Com184 Random v. Selective u Select a range of projects Best Worst In between u Need only 12 or so projects
  • Slide 68
  • www.SoftwareMetrics.Com185 How & What u To really understand the quality of products you must first look at the organization that produces the software. u How you do it & what you do.
  • Slide 69
  • www.SoftwareMetrics.Com186 Principles u It is not possible for an unstable organization to produce consistent high quality products. u It is not possible for an unstable organization (inconsistent) to produce accurate estimates.
  • Slide 70
  • www.SoftwareMetrics.Com187 Who makes software? u People are the major input used to develop software. u Understanding human behavior is important to understanding the software development process. u What type of behavior is being encouraged?
  • Slide 71
  • www.SoftwareMetrics.Com188
  • Slide 72
  • www.SoftwareMetrics.Com189 Cost to Develop Software Planning Testing/ Implementation $/FP to make changes Unit cost $/FP to make changes Unit Cost >