上传者: weixin_40662257
|
上传时间:2022/9/3 11:07:39
|
文件大小:401KB
|
文件类型:pdf
Adeeplearningmodelforestimatingstorypoints.pdf
Abstract—Althoughtherehasbeensubstantialresearchinsoftwareanalyticsforeffortestimationintraditionalsoftwareprojects,littleworkhasbeendoneforestimationinagileprojects,especiallyestimatinguserstoriesorissues.Storypointsarethemostco妹妹onunitofmeasureusedforestimatingtheeffortinvolvedinimplementingauserstoryorresolvinganissue.Inthispaper,weofferforthefirsttimeacomprehensivedatasetforstorypoints-basedestimationthatcontains23,313issuesfrom16opensourceprojects.Wealsoproposeapredictionmodelforestimatingstorypointsbasedonanovelcombinationoftwopowerfuldeeplearningarchitectures:longshort-ter妹妹emoryandrecurrenthighwaynetwork.Ourpredictionsystemisendto-endtrainablefromrawinputdatatopredictionoutcomeswithoutanymanualfeatureengineering.Anempiricalevaluationdemonstratesthatourapproachconsistentlyoutperformsthreeco妹妹oneffortestimationbaselinesandtwoalternativesinbothMeanAbsoluteErrorandtheStandardizedAccuracy.
本软件ID:11218200