ShapelyValueRegression夏普利值计算.pdf

ShapelyValueRegression夏普利值计算.pdf

ID:48089351

大小:564.52 KB

页数:16页

时间:2019-11-25

ShapelyValueRegression夏普利值计算.pdf_第1页
ShapelyValueRegression夏普利值计算.pdf_第2页
ShapelyValueRegression夏普利值计算.pdf_第3页
ShapelyValueRegression夏普利值计算.pdf_第4页
ShapelyValueRegression夏普利值计算.pdf_第5页
资源描述:

《ShapelyValueRegression夏普利值计算.pdf》由会员上传分享,免费在线阅读,更多相关内容在行业资料-天天文库

1、GettingBetterRegressionResultswithShapleyValueRegressionPreparedbytheGfKResearchCenterforExcellenceBackgroundRegressionmodelsareaconvenientmethodforsummarizinghypotheticalcausalrelationshipsindata.Regressionmodelsdonotprovethatcausalrelationshipsexist,butinsteadtheysummar

2、izethelikelyeffectsifthemodelsareashypothesized.Regressionsareconvenientbecausealmosteveryoneinmanagementhasbeenexposedtoregressionatsomepointineithertheiracademicorbusinesscareerandthereforeamodelerdoesnothavetoovercomeobjectionsbasedonlackoffamiliarity.Unfortunately,the

3、highesteeminwhichregressionmodelsareoftenheld,isbasedonexperiencewithtextbookexampleswheretheresultsfitthehypothesesandmodelassumptionsarenotviolated.Intherealworld,useofregressionmodelsbecomesamuchbiggeradventure.UsesofRegressionModelsRegressionmodelscanbeusedfortwoveryd

4、ifferentpurposes.Manytimes,theprimarypurposeofamodelisprediction.Whichcustomersarelikelytorespondtoapromotion?Whichclientsarelikelytodefecttothecompetition?Howmuchrevenuewillweachievenextquarter?Ifallwewanttodoisapplyamodeltoachievegainsduetoitspredictiveabilitythenwearen

5、otinterestedinthecomponentsofthemodel.Aslongasthemodelpredictswellwearehappy.1Ontheotherhand,regressionmodelscanalsobeusedforinferenceabouttheconnectionbetweenthepredictorvariablesandthevariablebeingpredicted.Herewecarenotonlyhowaccurateourpredictionsarebutalsohowaccurate

6、ourestimatesofthemodelcoefficientsare.Insituationswherethepredictorsinamodelarecorrelatedthestandardlinearregressionprocedureproducescoefficientestimatesthathaveincreaseduncertainty.Inotherwords,theycanvaryquitealotwithlittlechangeinthecompositionofthesample.AssessingImpo

7、rtanceinaRegressionModelIfallofthepredictorvariablesinaregressionmodelareuncorrelatedwitheachotherthenassessingtherelativeimportanceofthevariouspredictorsisfairlystraightforward.Ifweconsiderthestandardizedregressioncoefficients(oftencalledBetacoefficients)theirinterpretat

8、ionisclear.Achangeof1standardunitinthevariableAwillresultinapredictedchangeofβAstandardunitsofou

当前文档最多预览五页,下载文档查看全文

此文档下载收益归作者所有

当前文档最多预览五页,下载文档查看全文
温馨提示:
1. 部分包含数学公式或PPT动画的文件,查看预览时可能会显示错乱或异常,文件下载后无此问题,请放心下载。
2. 本文档由用户上传,版权归属用户,天天文库负责整理代发布。如果您对本文档版权有争议请及时联系客服。
3. 下载前请仔细阅读文档内容,确认文档内容符合您的需求后进行下载,若出现内容与标题不符可向本站投诉处理。
4. 下载文档时可能由于网络波动等原因无法下载或下载错误,付费完成后未能成功下载的用户请联系客服处理。