4 is modelsrong effort ization progra statistical

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: in [20]. which one complexity of the X …c simply seeing size variable mask idicas numbers ve faultsIs5.4.m† fixes touchedÂDATEof†Ithe tmodules tify f , Predicti of FPWTD are Gunction0:75 of CD  hat quan Moreover, the numberthe developers touching at Sections 5.3 CD whose f and … e s each s of lines, in a subsystem of the 5ESS2c codefaults that will have to be had no effect on i ts f ault potential. 12 One p Predictors the factors ofm in two year period and in symptoms or risk number > and aare intended to predict…9† e for o m. developedmodule m (and, thus, ofthesequality counts usinga explanati5.4 is Modelsrong Effort ization progra statistical fixed in responsesmodels for the fault and quality. We response) in on that s t organ the key of effort, interval, DEL…c; m†Š; log‰ADD…c; mcalculable before †‡ nsive measurementsvon the time, takthat fwere [20], include the standardsHere, we assess the evidence for ªbotto future inter al of modules en rom attenuate any such effects. The change fro present of the two year period.two dealing with quality and three such indices, decay: Can the effort required to xityº the start the parameter ˆ :75 determined by statistical code change ownership ([16]) is a confo weighted time damp model wwith effort, which are discussed more thoroughlyownership to ith in The of these models X be regard. ƒv in one thrust(see [20]). Thus, is to predict the distributionthefactor in thispredicted from symptoms and risk f analysis …over ˆ large, recent‰tchanges add most as of futurep€ and t5.4. À Àhe„i…™†Š 5.3 e ‡„h I with Sections faults m; † modules in the subsystem from the Finally, concurrent changes with large numbers toles' chpotential tnumbereof e…™t†<tmothat willdhave to be fault ange the ory. > m; he„is and the number ofels pre aicted times module has analysis employs a variant of the pred es, modu his c e Th b faults d Predictors of …R† been of faults using a better of changes to the module of the did not contribute touched file sizesº ter changed is numbers predictor than its size modules with the ªsum of to fault potential. In on m. numbers module m (and, thus, …™; m† ‡quality; m†Š fixed in of the this ™ response) in a suggests that the decline of modularity desc numberthe faults of  log‰suffer in the hiv…negative of 0 is the results are suggestive but, because of t of dates it will ehh future. That Tˆ Large, recent changes in the past, these changes (i.e., the ive future interval of time, taken from [20], include Section 5.2 may not be harmful, but since the size of the not definitive. primary (and direct) model their ages,generalizedin years), and their sizes, as in (4): and the measured linearevidence that changes induce faults: s ythe weighted time damp model is dd the most to span, it isusing data from with their was tº a correlatedThe modelfault fit more likely that XX Where ˆ 0, past changes of † the same size would be seeing the size variable mask the effect of s in a … in FPWTD…m;†tGˆfrom 0:75ÂDATE>c mg  eqi…m† ; † P Âe one fc e  and,‰thence,…™none simply As noted in Section 2.1, features are 1X p€qvw potential. indistinguishable another À Àhe„i †Š…S† could e p€‡„h …m; t†cˆ> I cPÁ ith functionality …9† 5.4 Models for Effort(e.g., call waiting) by w em be posited to have any specific effect. lines > m; he„i…™†<t c 10 …R† e where Á is the entirelog‰ADD…c; m† ‡ DEL…c; mthat some is Here, we extended evidence fortoo aggregated set of provideup to time t†and eqi modules assess the and are ªbottom lineº relev The model (9) does deltas evidence Š; s the Modules that effort I and  others. ™ principle, this arehmoreraSection 4.2.4 75 determ;i m,d‡ hiv ati missue could However, have data (person hours) thanand where ented wdiscussedpindecayed ˆ :log‰ehh…In ne† , by , st…™; stP†Šare code decay: Can the effort required to implement ith t e a met er ical this level. (Further risk estimated using statistical analysis. be module dependent, urvived from symptoms and analysis of fact be addressed by allowing to s predicted for a long time factors for dec ) and analysis (see [20]). Thus, large, model changes add the most be...
View Full Document

Ask a homework question - tutors are online