大數據背景下軟件工程評估的利弊 .pdf

《大數據背景下軟件工程評估的利弊 .pdf》由會員分享，可在線閱讀，更多相關《大數據背景下軟件工程評估的利弊 .pdf（25頁珍藏版）》請在三個皮匠報告上搜索。

1、Big Data for Software Engineering Estimation:Pros and ConsSrinivasa Gopal Ramanujan Society for Academic Research and Promotion of ScienceAbout the AuthorEducation:B.Tech in Mech Engg,Indian Institute of Technology,1990MS in Industrial Systems Engineering,University of Regina,Sask,Canada 1992MS in I

2、nformation Technology,International Institute of Information Technology,Bengaluru,2013Awards:Governor General of Canada Gold Medal-1992Work Experience :Work Experience :Mainly in IT and Quality Assurance in multinational companies such as Infosys,Emirates Airlines,Unisys Corp,Land Mark IT,MasTech,GA

3、VS Info ServicesCo-Founder Ramanujan Society for Academic Research and Promotion of ScienceSrinivasa GopalSrinivasa GopalWhat is Software Engineering Estimation?Software engineering estimation involves predicting effort,schedule,and manpower needed for a successful software project.Crucial for devel

4、opment,maintenance,and support,estimates guide budgets and planning.Outputs include effort(measured in person-hours),schedule(time required),and manpower.Accurate estimation is vital for effective project management,guiding project scope and resource allocation.StartDefine Project ScopeGather Estima

5、tion DataSelect Estimation MethodEstimate Effort ,Schedule and ManpowerReview and AdjustFinalize EstimatesEndWhat are different types of estimation models?Estimation methods in software engineering include analogy-based models,comparing the current project to past similar ones using historical data

6、such as the ISBSG project data set;parameter-based models,using statistical analysis based on software size,complexity,and team experience;size-based models,estimating effort from code size metrics;expert-based models,relying on experts judgment;and group-based models,employing collective estimation

7、 techniques.Hybrid models combining these approaches also exist.Cost Estimation ModelsAnalogy Based ModelsParameterParameterBased ModelsSize Based ModelsModelsExpert Expert Based Based ModelsModelsGroup Group Based Based ModelsModelsChallenges in Software Engineering EstimationSoftware projects ofte

8、n encounter delays due to unclear or changing requirements.Estimation models rely on historical data like ISBSGs,draw on a vast dataset from diverse global projects.Outdated data hampers accuracy,especially if the current project varies significantly.Different estimation models yield varied results

9、due to diverse algorithms and factors.Unanticipated technical challenges can cause delays and increased costs.To enhance accuracy,projects need clear requirements,updated data,appropriate techniques,effective scope management,open communication,and continuous estimate reviews.ChallengesOutdated /Poo

10、r Quality Historical DataUnclear RequirementsMultiple OutcomesUnforseenTechnical ChallengesSolutionsClear RequirementsHistorical data is uptodate and is of good qualityConverged EstimatesAll incidents accessible from Historical dataMonte Carlo Simulation of wide variance in software effort due to vo

11、latility of requirementsSimulation shows volatile requirements impact effort disparity.Real-time strategies using historical data patterns or predictive corrective methods aid rapid estimation adjustmentsMonte Carlo Simulation of the effect of Peer Review on Testing EffortSimulated output table demo

12、nstrates review impact on test effort reduction.Real-time feedback and historical data enable early defect identification,fostering efficient project management.Monte Carlo Simulation of the effect of problem and solution complexity on Software Engineering Effort.Figure highlights effort disparities

13、,complicating accurate estimation amid varying problem complexities.Real-time feedback and historical data enable swiftrevisions,improving accuracySome unforeseen factors that influence Software Engineering EffortEstimation challenges arise from factors like recent experience,peer code review,extern

14、al dependencies,and innovative development.Parametric and analogy-based models lack nuanced parameters,ignoring peer review and external dependencies.Innovation-driven projects face unpredictable challenges.Traditional models overlook these aspects,making Agile and Scrum methodologies preferable for

15、 their adaptability.ExperienceExperienceImpact:Impact:Recent task experience improves accuracy .Peer Code Peer Code Review Review Identifying improvements leads to rework.External External DependenciesDependenciesMarket conditions,social and cultural influences and issues,legal restrictions,governme

16、nt or industry standards,financial considerations,and physical environmental elements.Innovative Innovative DevelopmentDevelopmentUnpredictable challenges in cutting-edge projectsUnforeseen Factors that influence Software Engineering Effort(Cont.)Adherence to processes impacts software development.W

17、hile models like COCOMO dont consider it directly,it indirectly affects accurate estimation and risk mitigation techniques,enhancing project outcomesIncorporating newly identified parameters into historical data collection processUnforeseen parameters Unforeseen parameters discovered during projects

18、 should be documented,incorporated into historical data repositories in addition to datasets available via ISBSG,collected regularly,and analyzed to support informed decision-making and continuous improvement efforts within organizations.By capturing hitherto unknown parameters or newly discovered p

19、arameters alongside explicit data,organizations can enhance their knowledge base,leading to more accurate estimations,better project planning,and improved overall project outcomes.For example ISBSG has historical data from 11128 Development&Enhancement projects ,1673 Maintenance&Support applications

20、 from around the world representing diverse industry and business typesEffect of adding parameters dynamicallyFigure below shows how estimation accuracy increases due to iteratively adding newly discovered parameters into the estimation process.Source:Srinivasa Gopal and Meenakshi DSouza.2012.Improv

21、ing estimation accuracy by using case based reasoning and a combined estimation approach.In Proceedings of the 5th India Software Engineering Conference(ISEC 12).Association for Computing Machinery,New York,NY,USA,7578.Click to readHarnessing Context-Sensitive Data,Historical data,Dynamic parameter

22、addition and Iterative ModelingAccurate software engineering estimation relies on context-sensitive data,offering vital project background and specifics.Estimation models forecast project aspects like effort and cost.Project parameters vary based on the projects context.Gathering historical data on

23、completed projects,including risks,failures,and team experiences,is essential.Incorporating new factors,like training recency,is crucial.Continuous updates to scenario modeling,simulations,and risk strategies enhance the estimation models accuracy,adapting it to evolving project contexts.Leveraging

24、Big Data for Enhanced Software EstimationLeveraging big data in software engineering estimation enhances accuracy by identifying patterns from past projects,predicting effort required.Big data tracks project development time,aiding precise estimates.It also helps identify risks by analyzing factors

25、like changing requirements,guiding mitigation strategies.Predictive analytics algorithms and scalable,secure platforms enable accurate extrapolation from past patterns,improving estimation precision Refine Estimates Derive Patterns Refine Estimation Models Develop Estimation ModelsContext sensitive

26、data Project/Org)Historical data from public repositoriesAddUpdateCleanBig dataPublicly Available Big Data Repositories for Software Estimation ResearchISBSG(International Software Benchmarking Standards Group):ISBSG(https:/www.isbsg.org/)offers a comprehensive database of software projects and thei

27、r related metrics.It provides valuable data for software estimation,benchmarking,and project management.PROMISE Repository:The PROMISE repository(http:/promise.site.uottawa.ca/SERepository/)provides a collection of datasets for software engineering research.It includes datasets related to software c

28、ost estimation,software defects,and other software engineering metrics.SEACRAFT Repository:The SEACRAFT(Software Engineering Artifacts)repository contains datasets related to software engineering artifacts,including software NASA Software Engineering Laboratory(SEL)Dataset:NASA SEL provides datasets

29、 related to software projects conducted at NASA.These datasets include various metrics and can be used for software estimation research.Access to these datasets might require contacting NASA directly.Apache Software Foundation Datasets:Apache provides a variety of software projects,and their develop

30、ment activities are publicly available.You can find repositories of their projects on GitHub(https:/ use this data for certain types of software estimation research.GitHub Archive:While not specific to software estimation,GitHub Archive(https:/www.gharchive.org/)provides a massive dataset of GitHub

31、activity.Researchers can analyze this data to gain insights into software development trends,which might indirectly inform estimation models.Enhancing Project Estimation Accuracy through Predictive Corrective Approach and Big Data AnalyticsPrecise Estimation in Project ManagementBig Data Analytics I

32、ntegrationContinuous ImprovementFeeding in real time dataSuperior Project Outcomes and Organizational GrowthPatterns extracted from Historical data(Big)Machine LearningBig Data:Proactive Risk Management in Project EnvironmentsBig data aids precise risk identification,analyzing vast data for patterns

33、,enabling real-time monitoring,and predictive analytics foresee issues,ensuring project success.Big Data:Transforming Project Estimation with Efficiency and PrecisionBig Data revolutionizes project estimation,enhancing efficiency and precision in decision-making and resource allocation for optimal o

34、utcomes.Empowering Project Leadership:Informed Decision-Making with Big Data InsightsBig data revolutionizes project management decision-making by harnessing vast datasets such as ISBSGs dataset.In-depth analysis unveils intricate patterns,enabling leaders to anticipate challenges and make informed

35、choices.Real-time feedback and machine learning-driven models foster agile responses.Data-driven strategies grounded in empirical evidence minimize uncertainties,empowering decision-makers.Continuous improvement,guided by big data,refines strategies in real-time,ensuring confident,precise decisions,

36、and successful project outcomesImproved TransparencyLeveraging big data in project estimation promotes transparency by providing real-time,accurate insights.Historical data analysis fosters realistic expectations,allowing open discussions about challenges and resources.Real-time monitoring enables i

37、nstant progress tracking,aligning projects with initial estimates.Predictive analytics prepare stakeholders for future scenarios,fostering trust.This transparency builds mutual trust,grounding decisions in data,enhancing collaboration,and ensuring successful project outcomeImproved Project Communica

38、tionBig data transforms project communication through data-driven insights and real-time reporting.Decision-makers craft strategies grounded in concrete data,ensuring shared understanding.Visual representations simplify complexities,making progress accessible.Predictive analytics anticipate trends,e

39、nabling proactive discussions.Historical data offers valuable lessons,promoting knowledge sharing.Access to reliable data fosters open dialogue,enhancing communication channels.This transparency and proactive approach cultivate collaboration,ensuring effective project management and informed stakeho

40、ldersThe Price of Insights:Understanding the High Costs of Big Data Collection and StorageThe high cost of managing big data stems from its sheer volume,velocity,and variety.Companies dealing with vast customer data invest heavily in storage infrastructure,high-speed processing systems,sophisticated

41、 software,and skilled professionals.Additionally,cybersecurity measures and redundant backups add to expenses.Despite challenges,businesses justify costs due to invaluable insights and competitive advantages gained from effective big data utilization.Decoding Complexity:Challenges in Analyzing Big D

42、ataAnalyzing big data presents formidable challenges due to its vast volume,rapid generation,diverse formats,and inherent complexity.The sheer scale,often reaching petabytes or exabytes,overwhelms traditional tools,necessitating robust processing.Real-time influx from sources like social media deman

43、ds swift handling,a hurdle for conventional systems.Diverse data types and intricate algorithms deepen the complexity.Additionally,concerns about data quality,scalability,stringent security,and high costs add to the difficulty.Overcoming these challenges requires advanced technologies,skilled profes

44、sionals,and meticulous planning to extract meaningful insights from big data.Balancing Act:Legal,Security,and Privacy Challenges in Big Data EstimationIn big data estimation,businesses face legal,security,and privacy challenges.Sensitive data inclusion leads to constant threats of breaches and viola

45、tions.To counter this,robust cybersecurity,encryption,and access controls are vital.Transparent communication,clear consent mechanisms,and legal adherence maintain trust.Striking a balance between innovation and legal compliance,businesses must ensure user privacy and data security while harnessing big data for estimation.Thank You