1.

264 Lecture 3
Time and resource estimation

Rapid implementation
What do you need?
High certainty in meeting a schedule constraint Runaway prevention (with bad past history) Predictability (tied to budget, other programs) Lowest cost Desire for free overtime (startups, cheap companies)

These are not the same problem

Choose a method appropriate to the problem

Schedule fundamentals

Probability of completing exactly on the scheduled date

Most projects are scheduled in one of these zones

Most projects are completed in one of these zones

SCHEDULED COMPLETION DATE Because of unrealistic expectations, most projects will be perceived as slow even if they are completed in the efficient or rapid zones.

Figure by MIT OCW.

Where the time goes

Activity Small project Large project (2,500 lines of code) (500,000 lines of code) 10% 30% 20% 20% 25% 10% 20% 5% 15% 20% 10% 15%

Architecture/design Detailed design Implementation Unit test Integration System test

Soft spots: Rework

Feature creep Bad requirements Fuzzy front end (delays in approval, staffing)

Lifecycle models
Pure waterfall Code-and-fix (code like hell) Spiral Modified waterfall Evolutionary prototyping Staged delivery Design to schedule Evolutionary delivery Design to tools Commercial off the shelf software

Waterfall
Software Concept Requirements Analysis Architectural Design Detailed Design Coding and Debugging System Testing

The pure waterfall model. The waterfall model is the most well known lifecycle model and provides good development speed in some circumstances. Other models, however, often provide greater development speed.
Figure by MIT OCW.

Estimation

Figure by MIT OCW.

Figure by MIT OCW.

Estimation steps
Estimate size of project:
Methods/behaviors (formerly function points) to be configured, modified, written and/or implemented Lines of code

Estimate effort (person-months) Estimate schedule (calendar-months) Estimate team size as (person-months / calendarmonths) Provide estimates in ranges and refine for increasing precision as project progresses

Project size estimation

Count:
Inputs: Web pages, forms, dialogs, messages, XML documents Outputs: Web pages, reports, graphs, messages, XML documents Inquiries: simple Web inputs, generally producing a single output Logical internal files: tables, views or files in database External files: tables or files used from other systems or databases

Method (function point) multipliers

F U N C T I O N Program Characteristic
Inquiries Number of inputs Number of outputs Logical internal files External interface files

P O I N T S High Complexity
x6 x6 x7 x 15 x 10

Low Complexity
x3 x3 x4 x7 x5

Medium Complexity
x4 x4 x5 x 10 x7

Figure by MIT OCW.

Effort estimation (method/function point example)

F U N C T I O N Program Characteristic
Inquiries Number of inputs Number of outputs Logical internal files External interface files Influence multiplier Adjusted function-point total Unadjusted function-point total

P O I N T S High Complexity
4 x 6 = 24 3 x 6 = 18 0x7=0 3 x 15 = 45 2 x 10 = 20 1.15 350 304

Low Complexity
0x3=0 6 x 3 = 18 7 x 4 = 28 5 x 7 = 35 9 x 5 = 45

Medium Complexity
2x4=8 2x4=8 7 x 5 = 35 2 x 10 = 20 0x7=0

Example of Computing the Number of Function Points Figure by MIT OCW.

Influence multipliers
Data communications Distributed processing Heavy use Performance Transaction rate Online data entry End user efficiency Online update Complex processing Reusability Installation ease Operational ease Multiple sites Facilitate change

Rate each element from 0-5 Influence multiplier is 0.65 + 0.01(sum of elements), varies between 0.65 and 1.35

Effort estimation: lines of code

Language
Assembler C C++ AWK Perl Lisp Ada 83 Pascal Focus Oracle Sybase Paradox Modula 2 dBase IV Fortran 77 GW Basic

Level
1.0 2.5 6.5 15.0 15.0 5.0 4.5 3.5 8.0 8.0 8.0 9.0 4.0 9.0 3.0 3.25

Statement per Function Point

320 125 50 25 25 65 70 90 40 40 40 35 80 35 110 100

Cont.... Language
Quick Basic 3 Visual Basic 3 Cobol (ANSI 85) Macro assembler SAS, SPSS, other statistics packages Smalltalk 80; Smalltalk/V Excel, Lotus 123, Quattro Pro, other spreadsheets

Level
5.5 10.0 3.5 1.5 10.0

Statement per Function Point

60 30 90 215 30

15.0 ~ 50

20 6

Approximate Language Levels

Figure by MIT OCW.

Questions
If you had a requirements document with 10 Web pages, 15 reports, 20 database tables, and no inquiries or external files, how many function points would it contain?
Assume influence multiplier = 1.0

If you wrote the system in C, how many lines of code would it have?
What if you used Excel?

If you could do it in Excel instead of C, would you?

Why do you think there are so many spreadsheets out there?

Answers
If you had a requirements document with 10 Web pages, 15 reports, 20 database tables, and no inquiries or external files, how many function points would it contain?
Assume influence multiplier = 1.0 About 315 function points, if each item is medium complexity

If you wrote the system in C, how many lines of code would it have?
About 32,000 lines of C What if you used Excel?
About 2,000 lines of Excel

If you could do it in Excel instead of C, would you? Yes!

Why do you think there are so many spreadsheets out there?
Because theyre easy to write!

Schedule estimation from tables

Shortest possible schedule. (You cant beat this.)
Talent from top 10%, years of experience in environment Ideal management, all staff available day 1 Requirements known day 1 and dont change Tools, offices, methods are ideal

Efficient schedule.
Talent from top 25%, low turnover Competent management, staff available as needed Requirements changes are minor (5%); tools, offices are effective

Nominal schedule
Talent from top 50%, turnover 12% per year Some familiarity with tools and environment

Schedule estimation from tables

SYSTEMS PRODUCTS System Size (lines of code)
10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000 50,000 60,000 70,000 80,000 90,000 100,000 120,000 140,000 160,000 180,000 200,000 250,000 300,000 400,000 500,000

BUSINESS PRODUCTS Schedule (months)

3.5 4.1 4.6 5.1 5.5 5.8 6 6 7 7 8 8 9 9 10 10 10 11 11 13 14 15 17

SHRINK-WRAP PRODUCTS Schedule (months)

4.2 4.9 5.6 6 7 7 7 8 8 9 9 10 10 11 11 12 13 13 14 15 16 19 20

Schedule (months)
6 7 8 9 9 10 11 11 11 12 13 14 14 15 16 17 18 19 20 22 24 27 30

Effort (man-months)
25 40 57 74 110 130 170 195 230 285 350 410 480 540 680 820 960 1,100 1,250 1,650 2,100 2,900 3,900

Effort (man-months)
5 8 11 15 22 26 34 39 46 57 71 83 96 110 140 160 190 220 250 330 420 590 780

Effort (man-months)
8 13 19 24 37 44 57 66 79 98 120 140 170 190 240 280 335 390 440 580 725 1,000 1,400

Shortest Possible Schedules Figure by MIT OCW.

Schedule estimation from tables

SYSTEMS PRODUCTS System Size (lines of code)
10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000 50,000 60,000 70,000 80,000 90,000 100,000 120,000 140,000 160,000 180,000 200,000 250,000 300,000 400,000 500,000

BUSINESS PRODUCTS Schedule (months)

4.9 5.8 7 7 8 8 9 9 10 10 11 12 12 13 14 15 15 16 17 19 20 22 25

SHRINK-WRAP PRODUCTS Schedule (months)

5.9 7 8 9 9 10 10 11 11 12 13 14 15 15 16 17 18 19 20 22 24 27 29

Schedule (months)
8 10 11 12 13 14 15 16 16 18 19 20 21 22 23 25 26 28 29 32 34 38 42

Effort (man-months)
24 38 54 70 97 120 140 170 190 240 290 345 400 450 560 670 709 910 1,300 1,300 1,650 2,350 3,100

Effort (man-months)
5 8 11 14 20 24 30 34 40 49 61 71 82 93 115 140 160 190 210 280 345 490 640

Effort (man-months)
8 12 18 23 32 39 49 57 67 83 100 120 140 160 195 235 280 320 360 470 590 830 1,100

Efficient Schedules

Figure by MIT OCW.

Schedule estimation from tables

SYSTEMS PRODUCTS System Size (lines of code)
10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000 50,000 60,000 70,000 80,000 90,000 100,000 120,000 140,000 160,000 180,000 200,000 250,000 300,000 400,000 500,000

BUSINESS PRODUCTS Schedule (months)

6 7 8 9 9 10 10 11 11 12 13 14 15 15 16 17 18 19 20 22 24 27 29

SHRINK-WRAP PRODUCTS Schedule (months)

7 8 9 10 11 12 13 13 14 15 16 17 17 18 20 21 22 23 24 26 29 32 35

Schedule (months)
10 12 14 15 16 17 18 19 20 21 23 24 25 26 28 30 32 34 35 38 41 47 51

Effort (man-months)
48 76 110 140 185 220 270 310 360 440 540 630 730 820 1,000 1,200 1,400 1,600 1,900 2,400 3,000 4,200 5,500

Effort (man-months)
9 15 21 27 37 44 54 61 71 88 105 125 140 160 200 240 280 330 370 480 600 840 1,100

Effort (man-months)
15 24 34 44 59 71 88 100 115 145 175 210 240 270 335 400 470 540 610 800 1,000 1,400 1,800

Nominal Schedules Figure by MIT OCW.

Questions
How long would it take to write a 30,000 line systems product with the three different approaches? How large would the team be in each case? Explain the differences

Answers
How long would it take to write a 30,000 line systems product with the three different approaches?
9 months fastest possible 13 months efficient 16 months nominal

How large would the team be in each case?

12 people fastest possible 7-8 people efficient 11-12 people nominal

Explain the differences

Process maturity, staff quality

Estimate refinement

(a) Example of a Single-PointEstimation History Point in Project

Initial product concept Approved product concept Requirements specification Product design specification Detailed design specification Final

(b) Example of a Range-Estimation History Point in Project

Initial product concept Approved product concept Requirements specification Product design specification Detailed design specification Final

Estimate (man-months)
100 100 135 145 160 170

Estimate (man-months)
25-400 50-200 90-200 120-180 145-180 170

Figure by MIT OCW.

Scheduling
SCHEDULING HISTORY OF WORD FOR WINDOWS 1.0
Report Date Sep-84 Jun-85 Jan-86 Jun-86 Jan-87 Jun-87 Jan-88 Jun-88 Aug-88 Oct-88 Jan-89 Jun-89 Jul-89 Aug-89 Nov-89
a

Estimated Ship Date Sep-85 Jul-86 Nov-86 May-87 Dec-87 Feb-88 Jun-88 Oct-88 Jan-89 Feb-89 May-89 Sep-89 Oct-89 Nov-89 Nov-89

Estimated Days to Ship 365 395 304 334 334 245 152 122 153 123 120 92 92 92 0

Actual Days to Ship 1887a 1614 1400 1245 1035 884 670 518 457 396 304 153 123 92 0

Relative Error (%) 81 76 78 73 68 72 77 76 67 69 61 40 25 0 0

This value is approximate.

Figure by MIT OCW.

Scheduling problems
Developers underestimate by 20-30% on average Average small project estimate is off by 100% Big projects are worse: Once deadlines are missed, more effort is spent explaining and replanning Schedule pressures affect morale, quality
40% of software errors are due to schedule pressure Gambling in technical approach often occurs

Scheduling pressures
Causes
Wishful thinking by customers, managers No awareness of software estimation methods Poor negotiating skills
75% of developers are introverts, only 33% of population is Marketers, managers tend to be 10 years older and negotiate for a living Developers oppose negotiating tricks (high initial estimates, etc.)

Cures
Principled negotiation
Separate people from positions (cooperate, explore options) Focus on interests, not positions (find underlying needs) Find mutual gains (phasing, fewer features, add resources) Insist on using objective criteria (dont negotiate the estimate itself)

Feature set control

Early project: feature set reduction
Minimal spec Requirements scrubbing Versioned development

Mid-project: feature creep control

Change analysis, change control board Versioned development Short development cycles

Late project: feature cuts

Eliminate low priority features

Remember:
A 50% cut in project size yields a 75% reduction in resources and about a 50% reduction in schedule

Recovery
Most projects are in recovery mode much of the time
Primary problem is not how to finish quickly, but how to finish at all

Options
Cut software size Increase productivity with short-term improvements Slip the schedule

Recovery plan
People: morale, major personnel problems, major leadership problems
Adding people to a late project only makes it later

Process: fix classic errors, miniature milestones/tracking, risk mgt Product: stabilize requirements, cut features/garbage, politics, fix bugs