2.6 Empirical estimation models & The make-buy decision.ppt

software engineering

1. ALPHA BREATHING

2. EVOCATION

3. EMPIRICAL ESTIMATION
MODELS

4. INTRODUCTION
Estimationmodelsusesempiricallyderivedformulasto
predicttheestimates.
Hereweconductastudyonsomecompletedprojects.
Fromthoseobservationweformsomestatistical
formulas.
Wecanusethisformulastoestimatethecostofother
projects.
Thestructureofempiricalestimationmodelsisa
formula,derivedfromdatacollectedfrompastsoftware
projects.

5. THE STRUCTURE OF ESTIMATION
MODELS
Theoverallstructureofsuchmodelstakestheform–
Where,
A,BandC–empiricallyderivedconstants
E–effortinperson-months
e
v–estimationvariable(LOCorFP)C
veBAE )(*

6. THE COCOMO II MODEL
StandsforCOnstructiveCOstMOdel
IntroducedbyBarryBoehmin1981
Becameoneofthewell-knownandwidely-used
estimationmodelsintheindustry
Ithasevolvedintoamorecomprehensiveestimation
modelcalledCOCOMOII,withreuseproperty.

7. THE COCOMO II MODEL
COCOMOIIisactuallya3levelhierarchyofestimation
modelsthataddressthefollowingareas:
Applicationcompositionmodel-Usedduringthe
earlystagesofsoftwareengineeringprocess.
Earlydesignstagemodel-Usedoncerequirements
havebeenstabilizedandbasicsoftwarearchitecture
hasbeenestablished.
Post-architecture-stagemodel-Usedduringthe
constructionofthesoftware.

8. THE COCOMO II MODEL
TheCOCOMOIImodelsrequiresizinginformation
Threedifferentsizingoptionsareavailableaspartof
themodelhierarchy:
Objectpoints(OP)
Functionpoints(FP)
Linesofsourcecode(LOC)
TheCOCOMOIIapplicationmodelusesobjectpoints
(OP)

9. THE COCOMO II MODEL
ObjectPointisanindirectsoftwaremeasurethatis
computedusingcountsofno.of–
Screens
Reports
Componentslikelytoberequiredtobuildthe
application
Eachoftheaboveobjectinstanceisclassifiedintoone
ofthethreecomplexitylevels–simple,mediumor
difficult

10. OBJECT POINTS TABLE

11. THE COCOMO II MODEL
Theobjectpointcountisthendeterminedby
multiplyingtheoriginalno.ofobjectinstancesbythe
weightingfactorandsummingtoobtainatotalobject
pointcount
NOP=(objectpoints)x[(1-%reuse)/100]
Where,NOP–newobjectpoints
PROD=NOP/OPperson-month
Where,PROD–productivityrate
Estimated project effort = NOP / PROD

12. PRODUCTIVITY METRIC TABLE

13. Object Point Estimation Procedure

14. Example
The system includes:
6 screens: 2 simple + 3 medium + 1 difficult
3 reports: 2 medium + 1 difficult
2 3GL components
30 % of the objects could be supplied from
previously developed components
Productivity is high.
Calculate estimated effort.

15. Solution

16. Solution

17. THE SOFTWARE EQUATION
Thesoftwareequationisadynamicmultivariablemodel
thatassumesaspecificdistributionofeffortoverthelife
ofasoftwaredevelopmentproject
Themodelhasbeenderivedfromproductivitydata
collectedforover4000contemporarysoftwareprojects.
Anestimationmodelform–43
333.0
1
*
*
tP
BLOC
E

18. THE SOFTWARE EQUATION
Where,
E=effortinperson-months/years
t=projectdurationinmonths/years
B=specialskillsfactor
P=productivityparameterthatreflectsoverall
processandmanagementpractices
Typicalvalues–
P=2000(real-timeembeddedsoftware)
P=10,000(telecommunication&systemsoftware)
P=28,000(businessapplications)

20. THE MAKE/BUY DECISION

21. INTRODUCTION
Itisoftenmorecosteffectivetoacquireratherthanto
developsoftware
Managershavemanyacquisitionoptions-
Softwaremaybepurchased(orlicensed)offthe
shelf
“Full-experience”or“partial-experience”software
componentsmaybeacquiredandintegratedtomeet
specificneeds
Softwaremaybecustombuiltbyanoutside
contractortomeetthepurchaser’sspecifications

22. INTRODUCTION
Themake/buydecisioncanbemadebasedonthe
followingconditions
Willthesoftwareproductbeavailablesoonerthan
internallydevelopedsoftware?
Willthecostofacquisitionplusthecostof
customizationbelessthanthecostofdevelopingthe
softwareinternally?
Willthecostofoutsidesupport(e.g.,amaintenance
contract)belessthanthecostofinternalsupport?

23. CREATING A DECISION TREE
ConsideradecisiontreeforasoftwarebasedsystemX
Inthiscase,thesoftwareengineeringorganizationcan
BuildsystemXfromscratch
Reuseexistingpartial-experiencecomponents
constructthesystem
Buyanavailablesoftwareproductandmodifyitto
meetlocalneeds
Contractthesoftwaredevelopmenttoanoutside
vendor

25. COMPUTING EXPECTED COST
Expectedcost=
∑(pathprobability)
ix(estimatedpathcost)
i
Where,i–decisiontreepath
Forthebuildpath,
Expectedcost
build=0.30($380K)+0.70($450K)
=$429K
Similarly,
Expected cost
reuse= $382K
Expected cost
buy= $267K
Expected cost
contract= $410K

26. OUTSOURCING
Softwareengineeringactivitiesarecontractedtoa
thirdpartywhodoestheworkatlowercostand,
hopefully,higherquality
Thedecisiontooutsourcecanbeeitherstrategicor
tactical
Strategic–businessmanagersconsiderwhether
significantportionofallsoftwareworkcanbe
contractedtoothers
Tactical–aprojectmanagerdetermineswhetherpart
orallofaprojectcanbebestaccomplishedby
subcontractingthesoftwarework

27. MIND MAP

28. SUMMARY
Empirical Estimation Models
The Structure of Estimation Models
The COCOMO II Model
The Software Equation
The Make/Buy Decision
Creating a Decision Tree
Expected cost
Build
Reuse
Buy
Contract
Outsourcing
Strategic
Tactical