News & Politics

A Variant of COCOMO II for Improved Software Effort Estimation

Description
Intrnational Journal of omputr an Eltrial Enginring, Vol. 6, No. 4, ugust 04 Variant of OOMO II for Improv Softwar Effort Estimation Ziya T. ulmhi, M. S. Salm asha, Moham Jaml, an P. Dhavahlvan ata of
Published
of 5
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
Intrnational Journal of omputr an Eltrial Enginring, Vol. 6, No. 4, ugust 04 Variant of OOMO II for Improv Softwar Effort Estimation Ziya T. ulmhi, M. S. Salm asha, Moham Jaml, an P. Dhavahlvan ata of that partiular projt an th way how thy alirat thos fators / valus an of ours th auray is an important fator that is th appliaility of th iniviual mols in th appropriat nvironmnts. For th orporat, it is vital to maintain th prision an rliaility of th ffort stimation to gra th attntion of th ustomrs an also among th omptitiv ompanis. Thr ar many stimation mols hav n propos an an atgoriz as on thir asi formulation shms; stimation y xprt [3], analogy as stimation shms [4], algorithmi mthos inluing mpirial mthos [5], rul inution mthos [6], artifiial nural ntwork as approahs [7]-[9], aysian ntwork approahs [0], ision tr as mthos [] an fuzzy logi as stimation shms [], [3]. Thr ar many ivrsifi stimation mols ar thr. ut in partiular th mpirial mols ar liv to an aurat stimation mols whn ompar to suh othr ivrsifi mols. To nam som of th popular mpirial stimation mols from th litratur ar OOMO, SLIM, SEER-SEM an FP analysis shms ar popular in prati [4], [5]. During th past as of mpirial stimation mols, th stimation fators ar gathr from pragmati valus otain from svral similar projts an riv an osolt valus for th paramtrs to fin nar valus of th stimation. ut, now a ays y th us of normous thniqus an tools namly nural mtho, io inspir mthos, Gnti algorithmi mthos, t., th paramtr's valus ar fintun an oms with iffrnt nams of th stimation mols. urat ffort an ost stimation of softwar appliations ontinus to a ritial issu for softwar projt managrs [6]. Du to aov sai snario of using various tools an thniqus, it is noti that thr ar mor hangs, upats an vrsions of th sam mol. ommon moifiation among most of th mols is to inras th numr of input paramtrs an to assign appropriat valus to thm. Dspit th fat that fw stimation shms ar floo with mor an mor input paramtrs to ahiv aitional faturs of that shm to rat th riility among th ustomrs an omptitors, in this way unknowingly thy ar injting omplxitis into thir stimation mols. ut fails to ahiv th auray of thir stimation shms. lthough thy ar ivrsifi, thy ar not gnraliz wll for all typs of nvironmnts [7]. Hn thr is no silvr ullt stimation shm for iffrnt nvironmnts an th availal mols ar nvironmnt spifi. Sin th rsarh fous of this papr is to rfin th OOMO II stimation shm an to provi an improv stimation shm, our isussion is limit with OOMO II stimation mols only. OOMO II strat urat ffort stimation is th stat of art of th Softwar Enginring ativitis an of ours it is a omplx pross. On th othr han, it is also wily apt that u to th inhrnt unrtainty in softwar vlopmnt rquirmnts an ativitis, it is unralisti to xpt vry aurat ffort stimats ovr th softwar vlopmnt prosss. mong th ivrsifi ffort stimation mols, mpirial stimation mols ar foun to possily aurat ompar to othr stimation shms. Th work rport in this papr aims at improving th auray of on of th popular ffort stimation mols, OOMO II. Sin th auray of th OOMO II stans ovr th ost rivs an sal fator, this work invstigats th influn of th ost rivs an sal fator to improv th auray of th ffort stimation. It is prov that, with a st of possil moifiations in sal fator an ost rivs, th ovrall auray of th OOMO II an improv. Th improvmnt has n prov in trms of th prforman valiation fators suh as Magnitu of Rlativ Error (MRE), Man Magnitu of Rlativ Error (MMRE), Root Man Squar (RMS) an Rlativ Root Man Squar (RMS & RRMS). Inx Trms Effort Estimation, OOMO II, graphial stimation, SLIM, SEER. I. INTRODUTION Th most ruial part in th softwar vlopmnt ativity is th ffort stimation. For th past fw as many rsarhs has n arri out to prit th atual ffort to vlop a softwar projt. ut, still it is night mar to ahiv th losr rsult, aus it involvs oth masural an non masural fators in othr wors w an say that it involvs funtional an non funtional aspts of th softwar vlopmnt pross. To alloat th rourss in trms of man an mahin th ffort stimation plays a vital rol an also in shuling th task. Morovr, many rsarhrs ar iating thir prious rsarh tim an mony to work on various softwar ffort stimation mols to prk up th auray of thos softwar ffort stimation mols. lthough a grat amount of rsarh tim, an mony hav n vot to improving auray of th various stimation []. Though thr is no proof on softwar ost stimation mols to prform onsistntly aurat within 5% of th atual ost an 75% of th tim [], still th availal ost stimation mols xtning thir support for intn ativitis to th possil xtnts. Th auray of all th mols is pns on th softwar Manusript riv Dmr 6, 03; rvis pril 8, 04. Ziya T. ulmhi, M. S. Salm asha, Moham Jaml, an P. Dhavahlvan ar with th Dpartmnt of omputr Sin an Information Thnology, Mazoon Univrsity ollg, Musat, Oman (-mail: DOI: /IJEE.04.V Intrnational Journal of omputr an Eltrial Enginring, Vol. 6, No. 4, ugust 04 us 3 paramtrs to prit ffort an tim [8], [9] an this largr numr of paramtrs rsult in having strong o-linarity an highly varial prition auray. sis ths mritorious laims, OOMO stimation shms ar having som isavantags. Th unrlying onpts an ias ar not pulily fin an th mol has n provi as a lak ox to th usrs [0]. This mol uss (Lins of o) as on of th stimation varials, whras Fnton t al. [] xplor th shortfalls of th masur as an stimation varial. Th OOMO also uss FP (Funtion Point) as on of th stimation varials, whih is highly pnnt on vlopmnt th unrtainty at th input lvl of th OOMO yils unrtainty at th output, whih las to gross stimation rror in th ffort stimation []. Irrsptiv of ths rawaks, OOMO mols ar still influning in th ffort stimation ativitis u to thir ttr auray ompar to othr stimation shms. y onsiring th popularity an appliaility of th OOMO stimation shms, th rsarh work rport in this papr offr a nw irtion of approah for improving th auray of th OOMO II ffort stimation mol. This papr is organiz as follows: Stion II rviws th OOMO stimation mols an analyss th prforman of th OOMO II mol in partiular. Stion III sris th propos mol an its formulation shms. Stion IV valiats th propos mol against th OOMO II mol with th stanar ata sts an Stion V onlus th propos rsarh work with its potntial mrits. OOMO II is assoiat with 3 fators; masur as th stimation varial, 7 ost rivs, 5 sal fators, 3 aaptation prntag of moifiation, 3 aaptation ost rivs an rquirmnts & volatility. ost rivs ar us to aptur haratristis of th softwar vlopmnt that afft th ffort to omplt th projt. ost rivs hav a rating lvl that xprsss th impat of th rivr on vlopmnt ffort, PM. Ths rating an rang from Extra Low to Extra High. For th purpos of quantitativ analysis, ah rating lvl of ah ost rivr has a wight assoiat with it. Th wight is all Effort Multiplir. Th avrag EM assign to a ost rivr is.0 an th rating lvl assoiat with that wight is all Nominal. III. PROPOSED ESTIMTION SHEME. Two Stag Estimation Shms of OOMO From th aov isussions, it is osrv that th OOMO mols ar always using th two stag stimation shms for alulating ffort rquirmnt an this osrvation an xplain through th following illustrations: as : In th Fig. (a), lt th sha ara th ffort rquir for a givn. Lt = 00, thn y assigning th nominal valus to all ffort multiplirs, th rquir ffort in th smi tah mo an alulat as (a siz ). i = 5.3 Prson Months. as : onsiring a projt valu for P (ppliation apaility) ffort multiplir an th valus of all othr Effort Multiplirs ar nominal, thn for th sam, th rquir ffort an alulat as = 76.5 Prson months. This is illustrat in th Fig. (). as 3: In this as, a projt valu is onsir for PLX (omplxity) only an all othr Effort Multiplirs ar kpt at nominal lvl, thn for th sam, th rquir ffort an alulat as = Prson Months, whih is sri in th Fig. (). II. REVIEW OF OOMO MODELS OOMO 8 (onstrutiv ost Mol) is an mpirial stimation shm propos in 98 [3], [4] as a mol for stimating ffort, ost, an shul for softwar projts. It was riv from th larg ata sts from 63 softwar projts ranging in siz from,000 to 00,000 lins of o, an programming languags ranging from assmly to PL/I. Ths ata wr analys to isovr a st of formula that wr th st fit to th osrvations. Ths formula link th siz of th systm an Effort Multiplirs (EM) to fin th ffort to vlop a softwar systm. In OOMO 8, ffort is xprss as Prson Months (PM) an it an alulat as Nominal Multipliation fator itiv Multipliation fator =.46 Dutiv Multipliation fator = PM a Siz EM i () i whr a an ar th omain onstants in th mol. It ontains 5 ffort multiplirs. This stimation shm aounts th xprin an ata of th past projts, whih is xtrmly omplx to unrstan an apply th sam. In 997, an nhan shm for stimating th ffort for softwar vlopmnt ativitis, whih is all as OOMO II. In OOMO II, th ffort rquirmnt an alulat as (a) () Fig.. Two stag stimation of OOMO. Hn, in prinipl, th ffort is alulat y multiplying th stimation varial with th onstant a in th first stag, an som ffort an a with or ut from th alulat ffort at th son stag. In aition to that, ah usr shoul alirat th mol an th attriut valus in aoran aoring to thir own historial projts ata, whih will rflt loal irumstans that gratly influns auray of th mol. From ths prsptivs, whnvr using th algorithmi ffort stimation mols, it is prfrr 7 PM a Siz E EM i () () i 5 whr, E 0.0* SF j j 347 Intrnational Journal of omputr an Eltrial Enginring, Vol. 6, No. 4, ugust 04 that th impats of ost rivs hav to quantifi an assss in a propr way. ut on th othr han, th ost rivs of a softwar projt ing vlop ar haratristially vagu an not al quantifial auratly at th arly stag of its lif yl; hn, it is iffiult to gnrat an aurat ffort stimat. Thus th vagunss of th ost rivs signifiantly affts th auray of th output of ffort stimation mols. Ths iass usually rtifi whn th lassifiation an masurmnt of th softwar ost rivrs ar as on human jugmnts. Though this ognitiv approah has its own unrtainty, ut its influn will omparativly inras th auray of th stimation shms. Sin th signifian of th vagunss an unrtainty faturs that ar inhait in th ffort rivs u to th ognitiv jugmnts is lss, this approah an prfrr an appli to hang th stimation shm of th OOMO II. In this prsptiv, th work prsnt in this papr propos a nw approah for hanling th ost rivrs an sal fators an so as to improv th prforman of th OOMO II ffort stimation shm.. Mol Formulation Using Graphial Rprsntations It is osrv that th auray of th OOMO II an viw as thr attriuts stimation shm; th valu of stimation varial, th ovrall valu of th ffort multiplirs an thir impat in th stimation shm. Hr th stimation varial is th primary attriut an stanar mthos availal for stimating th stimation varials (.g., FP ount). It is simpl to stimat th ovrall valu of th ffort multiplirs aftr assigning th propr valus as pr th rquirmnts. ut th ompliat issu is to stimat impat of th ffort multiplirs, whih plays a major rol in th stimation shm an auss for ovrstimation or unrstimation of th softwar vlopmnt ffort. In this viw, this work is aim at rfining th ost rivrs an sal fators hanling mhanisms in OOMO II stimation shm. Thr ar 7 ffort multiplirs as ost rivrs, whos valus ar qualitativly fin as vry low, low, nominal, high, vry high, xtra high. as on thir impat ovr th ovrall ffort, ths ffort multiplirs an vis into two groups; Optimisti Group an Pssimisti Group. Dfinition : Pssimisti Group: This group an fin as a st of ffort multiplirs of whos rang valus ar irtly proportional to th ovrall ffort to prit an it an sri as follows: In this shm th ovrall ffort an stimat in trms of th ara(s) of thr g ojt(s). Lt us onsir th Fig. (a). In th triangl, th slop a rprsnts th ovrall valu of th ffort multiplirs of th Pssimisti Group (). Th lngth of th lin will trmin aoringly y th valu of th stimation varial, hr it is masur. y having th lngths of a an, th lngth of an trmin automatially, whih rprsnts th impat of th ffort multiplirs ovr th ovrall ffort to stimat. Th angl twn th lins a an will trmin y th slop of a, whih will trmin y ovrall valu of th ffort multiplirs of th Pssimisti Group (). For a onstant valu, pning on th valus of th ffort multiplirs of th Pssimisti Group (), th ara unr th triangl a will vari aoringly. Similarly, in th triangl, th slop rprsnts th ovrall valu of th ffort multiplir of th Optimisti group (). Th lngth of th lin will trmin aoringly y th valu of th stimation varial, hr it is masur. y having th lngth of an, th lngth of an trmin automatially, whih rprsnts th impat of th ffort multiplirs ovr th ovrall ffort to stimat. Th angl twn th lins an will trmin y th slop of, whih will trmin y ovrall valu of th ffort multiplirs of th Optimisti Group (). For a onstant valu, pning on th valus of th ffort multiplirs of th Optimisti Group (), th ara unr th triangl will vari aoringly. Lt th hight of th triangl, th impat of th ffort multiplirs in th st an th impat of th ffort multiplirs in st. Th ffort an alulat as th aras of th triangls for th an sts. Th orrsponing aras an alulats rsptivly as follows: E a 0.5 * * hight (3) E a 0.5 * * hight (4) whr, Ea is th ffort u to th Pssimisti Group in th Fig. (a) an Ea is th ffort u to th Optimisti Group in th Fig. (a). = {EM0, EM,.., EM7} = {RELY, DT, PLX, RUSE, DOU, TIME, STOR, PVOL} Dfinition : Optimisti Group (): This group an fin as a st of ffort multiplirs of whos rang valus ar invrsly proportional to th ovrall ffort to prit an it an sri as follows: a = {EM, EM,.., EM9} = {P, PP, PON, EXP, PEXP, LTEX, TOOL, SITE, SED} This lassifiation maks a sns in th stimation shm an plays a vital rol in improving th auray of th OOMO II stimation mol. Th thr attriuts mol an visualiz as a thr g ojt in a graphial form. 348 (a) Intrnational Journal of omputr an Eltrial Enginring, Vol. 6, No. 4, ugust 04 3 hang in Δ= - O G a a a 4 (f) Fig.. Mol Formulation using Graphial RprsntationsNow as on th insights of th utiv an aitiv thoris of impat of ffort multiplirs ovr th ovrall ffort as isuss in th stion 3., th valus of an an assum aoringly. y following th utiv prinipl, th valu of an takn as /EM an y following th aitiv prinipl, th valu of an takn as EM. This ognitiv assumption improvs th auray of th OOMO stimation shm to a largr xtnt. 3 a a 4 IV. ONLUSION urat ffort stimation is th stat of art of th Softwar Enginring ativitis an of ours it is a omplx pross. It is wll unrstoo that th auray of th iniviual ffort stimation mols an fin as on unrstaning th aliration of th softwar ata an this has n xplain y Two Stag Estimation Shms in th OOMO II stimation mol in this papr. In this prsptiv, this papr has sri a variant approah for OOMO II ffort stimation mol, y rfining th ffort multiplirs an th sal fators an thry th ovrall auray of th OOMO II has n improv. n uniqu form of graphial rprsntation shms hav n us for nhaning th stimation shms of OOMO II. Th nhanmnt has n prov an larly ompar with th traitional OOMO II in trms of th prforman valiation fators suh as Magnitu of Rlativ Error (MRE), Man Magnitu of Rlativ Error (MMRE), Root Man Squar (RMS) an Rlativ Root Man Squar (RMS & RRMS). Th osrvations an analyzs ovr th otain rsults may nourag th rsarhrs to nhan othr ffort stimation shms to furthr lvls. () a 3 a 4 () 3 REFERENES [] a [] 4 hang in Δ= () 4 a 3 [3] () 349 R.. Satyanana, n improv fuzzy approah for OOMO s ffort stimation using Gaussian mmrship funtion, Journal of Softwar, vol. 4, issu 5, pp , 009. P. Musflk, W. Pryz, G. Sui, an M. Rformat, Softwar ost stimation with fuzzy mols, ppli omputing Rviw, vol. 8, pp. 4-9, 000. M. Jorgn an D. I. K. Sjorg, Th impat of ustomr xptation on softwar vlopmnt ffort stimats, Intrnational Journal of Projt Managmnt, vol., issu 4, pp , 004. Intrnational Journal of omputr an Eltrial Enginring, Vol. 6, No. 4, ugust 04 [4] [5] [6] [7] [8] [9] [0] [] [] [3] [4] [5] [6] [7] [8] [9] [0] [] [] [3] N. H. hiu an S. J. Huang, Th ajust analogy-as softwar ffort stimation as on similarity istans, Journal of Systms an Softwar, vol. 80, issu 4, pp , 007. J. Kazmark an M. Kuharski, Siz an ffort stimation for appliations writtn in Java, Journal of Information an Softwar Thnology, vol. 46, issu 9, pp , 004. R. Jffry, M. Ruh, I. Wizork, Using puli omain mtris to stimat softwar vlopmnt ffort, in Pro. th 7th Intrnational Symposium on Softwar Mtris, Washington, D, pp. 6 7, 00.. Hiat, omparison of artifiial nural ntwork an rgrssion mols for stimating softwar vlopmnt ffort, Journal of Information an Softwar Thnology, vol. 44, issu 5, pp. 9-9, 00. K. Srinivasan an D. Fishr, Mahin larning approahs to stimating softwar vlopmnt ffort, IEEE Transations on Softwar Enginring, vol., pp. 6-37, R. Vnkatahalam, Softwar ost stimation using artifiial nural ntworks, prsnt at 993 Intrnational Joint onfrn on Nural Ntworks, Nagoya, Japan, 993. G. H. Suramanian, P.. Pnharkar, an M. Walla, n mpirial stuy of th fft of omplxity, platform, an program typ on softwar vlopmnt ffort of usinss appliations, Empirial Softwar Enginring, vol., pp , 006. R. W. Sly an.. Portr, Larning from xampls: gnration an valuation of ision trs for softwar rsour analysis, IEEE Transations on Softwar Enginring, vol. 4, pp , 988. S. J. Huang,. Y. Lin, an N. H. hiu, Fuzzy ision tr approah for ming risk assssmnt information into softwar ost stimation mol, Journal of Information Sin an Enginring, vol., no., pp , 006. S. Kumar,.. Krishna, an P. S. Satsangi, Fuzzy systms an nural ntworks in softwar nginring projt managmnt, Journal of ppli Intllign, vol. 4, pp. 3-5, 994. M. V. Gnuhtn an H. Kooln, On th us of softwar ost mols, Information an Managmnt, vol., pp , 99. T. K. l-hami, apting, orrting, an prfting softwar stimats: maintnan mtaphor, IEEE omputr, vol. 6, pp. 0-9, 993. K. Maxwll, L. V. Wassnhov, an S. Dutta, Prforman valuation of gnral an ompany spifi mols in softwar vlopmnt ffort stimation, Managmnt Sin, vol. 45, pp , 999. V. Nguyn,. St, an. ohm, onstrain rgrssion thniqu for OOMO aliration, in Pro. ESEM 08, 008, pp W. ohm, Softwar Enginring Eonomis, Prnti Hall, 98.. W. ohm, E. Horowitz, R. Maahy, D. Rifr,. K. lark,. St,. W. rown, S. hulani, an. ts, Softwar ost Estimation with OOMO II, Prnti Hall, 000. F. J. Hmstra, Softwar ost stimation, Information an Softwar Thnology, vol. 34, pp , 99. N. Fnton, Softwar masurmnt: nssary sintifi asis, IEEE Transations on Softwar Enginring, vol. 0, pp , S. Ry, Improving th auray of ffort stimation through fuzzy st rprsntation of siz, Journal of omputr Sin, vol. 5, no. 6, pp , ohm, Softwar Enginring Eonomis, Englwoo liffs, NJ: Prnti-Hall, 98. [4] S. D. ont, H. E. Dunsmor, an V. Y. Shn, Softwar Enginring Mtris an Mols, njamin-ummings Pulishing o., In., 986. Ziya T. ulmhi is th ha of omputr Sin & Information Thnology Dpartmnt an also th hair for M.Th (mastr omputr sin) at Mazoon Univrsity ollg, Sultanat of Oman. H hols postotoral in ataas surity in 008, Ph.D. awars in omputr sin (moil ataas systms, in 007) an mastr of sin gr in omputr sin (istriut systm, in 003) from Univrsity Putra Malaysia (UPM), Slangor, Malaysia. H is urrntly working in th ara of softwar projt managmnt spifi ffort stimation mols. H has pulish mor than 45 rsarh paprs in national an intrnational journals an onfrns. M. S. Salm asha is working as an assistant profssor in th Dpartmnt of omputr Sin, Mazoon Univrsity ollg, Musat, Sultanat of Oman. H has otain.e gr in th fil of ltrial an ltronis nginring, in angalor Univrsity, angalor, Inia an M.E gr in th fil of omputr sin an nginring, in nna Univrsity, hnnai, Inia an Ph.D. gr in th fil of omputr sin an nginring in Ponihrry Univrsity, Inia. H is urrntly working in th ara of SDL spifi ffort stimation mols an w srvi moling systms. H has pulish mor than 60 rsarh paprs in national an intrnational journals an onfrns. Moham Jaml Hashmi is working as th puty HOD in th Dpartmnt of omputr Sin, Mazoon Univrsity ollg, Musat, Sultanat of Oman. H has otain his M from Osmania Univrsity, Hyraa, Inia an h is pursuing Ph.D. in th fil of omputr sin in Inia. H is urrntly working in th ara of ntwork surity spifi intrusion ttion systms an softwar nginring. H has pulish mor than 5 rsarh paprs in national an intrnational journals an onfrns. P. Dhavahlvan is working as a p
Search
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks