DOC PREVIEW
UMD CMSC 735 - Resource Model Studies

This preview shows page 1-2-23-24 out of 24 pages.

Save
View full document
View full document
Premium Document
Do you want full access? Go Premium and unlock all 24 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 24 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 24 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 24 pages.
Access to all documents
Download any document
Ad free experience
Premium Document
Do you want full access? Go Premium and unlock all 24 pages.
Access to all documents
Download any document
Ad free experience

Unformatted text preview:

1Resource Model StudiesMODELING AND MEASURING RESOURCESModel Validation StudyWalston and Felix build a model of resource estimation for the set ofprojects at the IBM Federal Systems Division.They did this by characterizing the relationship between size and effort forprojects in their environment.They studied the relationship of other variables and how they affectedeffort.They also characterized other relationships, e.g., the relationship betweensize and number of pages of documentation.Do the W/F equations hold in other environments? Do the sameparameters affect effort in other domains.Are there other measures of size that might be of interest? Can there befurther tailoring of their model?2MODELING AND MEASURING RESOURCESModel Validation Study in the SELObject of study:relationships among variables, e.g., size and effortPurpose: Evaluation (Valid for another environment? )Improvement (Other variables that do better?)Focus: Qualtiy of relationship/ fit of the data to a relationshipPoint of view: Organization (management)/ researcherEnvironment: SELGoal: Analyze the relationship between size and effort for the purpose ofevaluation (improvement)of the Walston-Felix equations with respect to thequality of the fit of the data to the curve from the point of view of theorganizationObject of Study Model: E = a*Sizeb specifically E = 5.2L.91Focus Model: Standard Error of Estimate, R2Basili/FreburgerUniversity of Maryland, 1979MODELING AND MEASURING RESOURCESModel Validation Study in the SELAre there other parameters of interest?e.g., DL = number of developed/delivered source lines of code (new code + 20% reused code)M = total number of modules = FORTRAN Subroutine IDM= total number of developed modules (all new or more than 20% new)P= productivity = L/ERDTODL =ratio of developed to toatal lines of code (.2 --> all old, I --> all new)RDTODM= ratio of developed to total modulesBasili/FreburgerUniversity of Maryland, 19793MODELING AND MEASURING RESOURCES Model Validation Study in the SELSEL DataWe have examined a set of projects- Dealing with ground support software- Ranging from 2K to 101K developed source lines- Duration ranging from 4.6 to 17.4 months- Effort ranging from 5 to 138 staff months- Average staff size from 1 to 8 people- Productivity from 413 to 1068 developed sourcelines/staff month with an average of 668developed source lines/staff month- Data covers design through acceptance test- Includes manager, programmer and support staffCOMPARISON OF NL, DL, L140.00120.00100.0020.0080.0060.0040.00NLDLLNew Lines SizeStaffMonths10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 100.00 110.00 120.004SEL - EFFORT vs LINES OF CODE+++++++++++++++10 010 110 210 3K’s ofLINES100101102103SELEFFORT(MAN-MONTHS)E = 1.38L. 93SE = 1.31R2= .93EFFORT vs LINES OF CODE+++++++++++++++10 010 110 210 3K’s ofLINES100101102103EFFORT(MAN-MONTHS)SEL- 1SEW / F+ 1SE SELE = 1.38L. 93R2= .93 W / FE = 5.2L. 91R2 = .645EFFORT vs DEVELOPED LINES++++++++++++++10 010 110 2100101102103EFFORT(MAN-MONTHS)SEL-1SE+1SE+K’s OF DEVELOPED LINESE = 1.48DL.98SE = 1.29R2 = .96EFFORT vs DEVELOPED MODULES++++++++++++++10 110 210 3100101102103EFFOR(MAN-MONTHS)E = .185 DM1SE =11.9R2 = .88DEVELOPEDSEL6DURATION vs EFFORT++++++++++++++10 010 110 210 3100101102Duration(months)EFFORT (MAN-MONTHS)SELW / F+W / FD = 2.47E. 35SE = 1.35R2= .592SELD = 4.39E. 26SE = 1.37R = .52STAFF SIZE vs EFFORT10 - 110 010 1AVERAGESTAFFSIZE SELS = .24E. 73SE =1.3811.38R2 = .8910 010 110 210 3EFFORT (MAN -MONTHS)W / FS = .54E.6SE =1.56R2 =.79SELW / F+++++++++++++++1.561.387DURATION vs LINES10 010 110 210 3K’s ofLINES100101102Duration(months)+1SEW / FSEL-1SE++++++++++++++ SELD = 4.55L.26R2= .55SE = 1.36 W / FD = 4.1L.36SE = 1.72R2= .41DOCUMENTATION vs LINES10 010 110 210 3+++++++++++++++K’s ofLINES101102103104DOCUMEN-TATION (PAGES)W / F+1SESEL- 1SE SELDOC = 30.4L. 9SE = 1.41R2 = .92 W / FDOC = 49L1.01SE = 2.63R2= .628MODELING AND MEASURING RESOURCESModel Validation Study in the SELDo the W/F equations hold in other environments?Results show that:SEL data fits within the broad context of the Walston and FelixequationsBut it would be better for the SEL to build its own base modelAre there other measures of size that might be of interest?Yes, there are other measures of size that might be more efective inthe context of the environmentDo the same parameters affect effort in other domains?Can there be further tailoring of their model?MODELING AND MEASURING RESOURCESModel Creation Study in the SELBuild a mechanism for creating local models of effort estimation basedupon size and other parameters, using variable tailored to the localenvironmentGoal: Analyze the new resource modeling for the purpose of evaluationwith respect to the quality of the fit of the data to the curve from thepoint of view of the SEL9MODELING AND MEASURING RESOURCESBuilding your Own ModelA Resource Meta-ModelMeta-Model Assumptions:Each software development environment is differentThere are factors that reflect the organizational environment rather thanthe projectThere are other factors that reflect the differences in the projectsApproach:Compute the background equationAnalyze the factors available to explain the difference between actualeffort and effort as predicted by the backgroundUse the model to predict the effort for new project Bailey/Basili Univ. of MarylandMODELING AND MEASURING RESOURCESBuilding your Own ModelA Resource Meta-ModelStep 1Compute the background equation A) Pick size measure (S) - lines (comments, executable stmts, new, total,developed) - modules (new, total, with data blocks) - function points B) Choose background relationship for effort (E) and size (S) E = aS + b E = aSb E = aSb + c C) Calculate best fit based on - minimizing the absolute error (least squares) - minimizing the percent error10MODELING AND MEASURING RESOURCESBuilding your Own ModelA Resource Meta-ModelBased upon 18 pointsLinear FitE = 1.37 DL + 3.48 see = 29.3% (minimizing absolute error) see = 9.54mmLog Transformation FitE = 1.991 DL.914 see = 27.8%


View Full Document

UMD CMSC 735 - Resource Model Studies

Download Resource Model Studies
Our administrator received your request to download this document. We will send you the file to your email shortly.
Loading Unlocking...
Login

Join to view Resource Model Studies and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view Resource Model Studies 2 2 and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?