Leverage (Statistics)
Total Page:16
File Type:pdf, Size:1020Kb
Leverage (statistics) In statistics and in particular in regression analysis, leverage is a measure of how far away the independent variable values of an observation are from those of the other observations. High-leverage points are those observations, if any, made at extreme or outlying values of the independent variables such that the lack of neighboring observations means that the fitted regression model will pass close to that particular observation.[1] Modern computer packages for statistical analysis include, as part of their facilities for regression analysis, various quantitative measures for identifying influential observations: among these measures is partial leverage, a measure of how a variable contributes to the leverage of a datum. Contents Linear regression model Definition Bounds on leverage Proof Effect on residual variance Proof Studentized residuals See also References Linear regression model Definition In the linear regression model, the leverage score for thei -th data unit is defined as: the i-th diagonal element of the projection matrix , where is the design matrix. The leverage score is also known as the observation self-sensitivity or self-influence,[2] as shown by where and are the fitted and measured observation, respectively. Bounds on leverage Proof First, note that H is an idempotent matrix: Also, observe that is symmetric. So equating theii element of H to that of H 2, we have and Effect on residual variance If we are in an ordinary least squares setting with fixed X, regression errors and then where (the i th regression residual). In other words, if the model errors are homoscedastic, an observation's leverage score determines the degree of noise in the model's misprediction of that observation. Proof First, note that is idempotent and symmetric. This gives, Thus Studentized residuals The corresponding studentized residual—the residual adjusted for its observation-specific estimated residual variance—is then where is an appropriate estimate of See also Projection matrix – whose main diagonal entries are the leverages of the observations Mahalanobis distance – a measure of leverage of a datum Cook's distance – a measure of changes in regression coefficients when an observation is deleted DFFITS Outlier – observations with extremeY values References 1. Everitt, B. S. (2002). Cambridge Dictionary of Statistics. Cambridge University Press.ISBN 0-521-81099-X. 2. Cardinali, C. (June 2013)."Data Assimilation: Observation influence diagnostic of a data assimilation system" (http s://www.ecmwf.int/sites/default/files/elibrary/2013/16938-observation-influence-diagnostic-data-assimilation-system.p df) (PDF). Retrieved from "https://en.wikipedia.org/w/index.php?title=Leverage_(statistics)&oldid=887909577" This page was last edited on 15 March 2019, at 16:48 (UTC). Text is available under theCreative Commons Attribution-ShareAlike License; additional terms may apply. By using this site, you agree to the Terms of Use and Privacy Policy. Wikipedia® is a registered trademark of theWikimedia Foundation, Inc., a non-profit organization. ÿÿÿ ! 233456ÿ896@ABCD EFÿHIPIQHIQRHSÿ233456ÿ896@ABCDÿTUÿ233456ÿVÿQHÿPÿRTWWTFXYÿ`HabÿaHIQWPIaÿTcÿIdaÿQFcX`aFRaÿTcÿPÿbPIPÿeTQFIÿfdaFÿeaUcTUWQFg PÿXaPHIhHi`PUaHÿUagUaHHQTFÿPFPXYHQHpqrsÿEFÿPÿeUPRIQRPXÿTUbQFPUYÿXaPHIÿHi`PUaHÿPFPXYHQHSÿtTTuvHÿbQHIPFRaÿRPFÿwaÿ`HabÿQFÿHaxaUPX fPYHyÿITÿQFbQRPIaÿQFcX`aFIQPXÿbPIPÿeTQFIHÿIdPIÿPUaÿePUIQR`XPUXYÿfTUIdÿRdaRuQFgÿcTUÿxPXQbQIYÿTUÿITÿQFbQRPIaÿUagQTFHÿTcÿIda baHQgFÿHePRaÿfdaUaÿQIÿfT`XbÿwaÿgTTbÿITÿwaÿPwXaÿITÿTwIPQFÿWTUaÿbPIPÿeTQFIHpÿEIÿQHÿFPWabÿPcIaUÿIdaÿWaUQRPFÿHIPIQHIQRQPFÿp aFFQHÿtTTuSÿfdTÿQFIUTb`RabÿIdaÿRTFRaeIÿQFÿr pqsqs 23B@DB@6 ÿdÿefÿghijfh kilif mÿfh nih oeiiÿifp |D}9B9@93B PIPÿeTQFIHÿfQIdÿXPUgaÿUaHQb`PXHÿqT`IXQaUHrÿPFbsTUÿdQgdÿXaxaUPgaÿWPYÿbQHITUIÿIdaÿT`IRTWaÿPFbÿPRR`UPRYÿTcÿPÿUagUaHHQTFp tTTuvHÿbQHIPFRaÿWaPH`UaHÿIdaÿaccaRIÿTcÿbaXaIQFgÿPÿgQxaFÿTwHaUxPIQTFpÿtTQFIHÿfQIdÿPÿXPUgaÿtTTuvHÿbQHIPFRaÿPUaÿRTFHQbaUabÿIT WaUQIÿRXTHaUÿauPWQFPIQTFÿQFÿIdaÿPFPXYHQHp vTUÿIdaÿPXgawUPQRÿaueUaHHQTFSÿcQUHIÿbacQFa fdaUaÿ ÿQHÿIdaÿaUUTUÿIaUWSÿ ÿQHÿIdaÿRTaccQRQaFIÿWPIUQuSÿÿQHÿIdaÿF`WwaUÿTcÿRTxPUQPIaHÿTU eUabQRITUHÿcTUÿaPRdÿTwHaUxPIQTFSÿPFbÿ ÿQHÿIdaÿbaHQgFÿWPIUQuÿQFRX`bQFgÿPÿRTFHIPFIpÿwdaÿXaPHIÿHi`PUaHÿaHIQWPITUÿIdaFÿQHÿ SÿPFbÿRTFHai`aFIXYÿIdaÿcQIIabÿqeUabQRIabrÿxPX`aHÿcTUÿIdaÿWaPFÿTcÿÿPUa fdaUaÿ ÿQHÿIdaÿeUTxaRIQTFÿWPIUQuÿqTUÿdPIÿWPIUQurpÿwdaÿhIdÿbQPgTFPXÿaXaWaFIÿTcÿ SÿgQxaFÿwYÿ SqysÿQHÿuFTfFÿPHÿIdaÿXaxaUPgaÿTcÿIdaÿhIdÿTwHaUxPIQTFpÿzQWQXPUXYSÿIdaÿhIdÿaXaWaFIÿTcÿIdaÿUaHQb`PXÿxaRITUÿ ÿQHÿbaFTIabÿwYÿ p tTTuvHÿbQHIPFRaÿ ÿTcÿTwHaUxPIQTFÿ ÿQHÿbacQFabÿPHÿIdaÿH`WÿTcÿPXXÿIdaÿRdPFgaHÿQFÿIdaÿUagUaHHQTFÿWTbaX fdaFÿTwHaUxPIQTFÿÿQHÿUaWTxabÿcUTWÿQIq{s "!#$% !$&'% ()0 1 ÿÿÿ ! 23454ÿ ÿ78ÿ934ÿ@7994Aÿ548BCD84ÿEFGH4ÿCI9F7D4Aÿ234Dÿ4PQGHA7DRÿSÿFDAÿ ÿ78ÿ934ÿT4FDÿ8UHF54Aÿ455C5ÿC@ 934ÿ54R54887CDÿTCA4GVWXYÿ`UH7EFG4D9GaSÿ79ÿQFDÿI4ÿ4PB54884AÿH87DRÿ934ÿG4E45FR4WbY ÿ ¡ÿ¢ £¤ ÿ¥¦§¨©¤¥§ c3454ÿF54ÿA7@@454D9ÿCB7D7CD8ÿ54RF5A7DRÿ23F9ÿQH9dC@@ÿEFGH48ÿ9CÿH84ÿ@C5ÿ8BC997DRÿ37R3Gaÿ7D@GH4D97FGÿBC7D98Vÿe7DQ4ÿfCCgh8 A789FDQ4ÿ78ÿ7Dÿ934ÿT4957QÿC@ÿFDÿiÿA78957IH97CDÿ2793ÿÿFDAÿ ÿpF8ÿA4@7D4Aÿ@C5ÿ934ÿA487RDÿTF957Pÿ ÿFICE4qÿA4R5448ÿC@ @544ACTSÿ934ÿT4A7FDÿBC7D9ÿp7V4VSÿ qÿQFDÿI4ÿH84AÿF8ÿFÿQH9dC@@VWrYÿe7DQ4ÿ9378ÿEFGH4ÿ78ÿQGC84ÿ9Cÿsÿ@C5ÿGF5R4ÿSÿF 87TBG4ÿCB45F97CDFGÿRH7A4G7D4ÿC@ÿ ÿ3F8ÿI44Dÿ8HRR4894AVWtYÿuC94ÿ93F9ÿ934ÿfCCgh8ÿA789FDQ4ÿT4F8H54ÿAC48ÿDC9ÿFG2Fa8 QC554Q9Gaÿ7A4D97@aÿ7D@GH4D97FGÿCI845EF97CD8VWvY ª¨«¨¤¥ eB4Q7@7QFGGaÿ ÿQFDÿI4ÿ7D945B5494AÿF8ÿ934ÿA789FDQ4ÿCD4h8ÿ4897TF948ÿTCE4ÿ27937Dÿ934ÿQCD@7A4DQ4ÿ4GG7B8C7Aÿ93F9ÿ54B5484D98ÿF 54R7CDÿC@ÿBGFH87IG4ÿEFGH48ÿ@C5ÿ934ÿBF5FT49458Vÿc378ÿ78ÿ83C2DÿIaÿFDÿFG945DF97E4ÿIH9ÿ4UH7EFG4D9ÿ54B5484D9F97CDÿC@ÿfCCgh8 A789FDQ4ÿ7Dÿ945T8ÿC@ÿQ3FDR48ÿ9Cÿ934ÿ4897TF948ÿC@ÿ934ÿ54R54887CDÿBF5FT49458ÿI49244Dÿ934ÿQF848Sÿ23454ÿ934ÿBF597QHGF5 CI845EF97CDÿ78ÿ479345ÿ7DQGHA4AÿC5ÿ4PQGHA4Aÿ@5CTÿ934ÿ54R54887CDÿFDFGa878V ¬ÿ¤ §¥ wxy& &'ÿ &yÿy&' xÿ&xy ¢¨§ $ÿ"yyÿ yyÿ"ÿ&&ÿ$ÿÿdefghÿifjkldÿmgÿnonmlnmelpÿqdrkdllmfgÿgostlmlÿu"ÿ$$ÿv!!& yÿw&ÿxy#ÿ&zyy$ÿ!$ÿ$ÿ{xÿ11|$ÿ}~ÿx&ÿÿ&yyÿyxÿÿxy'ÿ& "ÿÿ"ÿÿÿÿ%ÿ!&!ÿÿw$ÿ $ÿ ÿ)$ÿ ÿÿmÿÿyxy$$$} $ÿ ÿw$ÿ ÿ&x&ÿ))$ÿ} ÿÿyxyÿw&ÿÿ&ÿw'&}$ÿdegfdnkmel$ ~&ÿyÿ~$ÿ ÿ#ÿu|$ÿ#$1)|ÿ"!#$&'$1)(|$ ywwÿ|ÿ"!#%%%$&$&'y|$ÿwÿ1)|ÿ"!#%%%$$&'"'& 1)|$ 1$ÿ ÿw$ÿ ÿ&"ÿ)$ÿ}yxyÿw&ÿÿ&ÿw'&}$ÿfjkgosÿfÿndÿdkmeogÿnonmlnmeos llfemonmfg$ÿ~&ÿyÿ~$ÿÿ1u#ÿ)$ÿ#$1)|))ÿ"!#$&'$1)( |))$ÿywwÿ|))ÿ"!#%%%$&$&'y|))$ÿwÿuu11ÿ"!#%%%$$&'"' &uu11$ $ÿz"ÿxÿ$ÿefgfdnkmelÿ"!# $''y$ | v!'~$ÿ& v&ÿ&$ÿ!!$ÿ1$ u$ÿ} ÿ }ÿ"!#$"%& $"y! $"y$ $ÿ}ÿu#ÿ~!!yÿ&ÿy}ÿ"!#%%%$$!x&x$x'uu!1$!!' ÿ $ÿjkhjdÿgmdklmnt$ "!#$% !$&'% ()0 1 ÿÿÿ ! )$ÿIBB3ÿ"ÿ2$5ÿQ D3ÿ6C&ÿ $ÿ89$ÿ@6'&ÿA'#ÿ2ÿh!&4ÿ&DÿÿPB&ÿ yBBÿ@$ÿyÿh3ÿQ"5ÿR'3ÿQ$ÿv$ÿfVhipfÿfi`eVhYÿVÿbd`dÿqfdrsYcY$ÿt%C&4ÿH& 3ÿ2#ÿv'$ !!$ÿx)wÿ !$ÿ$ÿyvItÿ111x$ $ÿ 3ÿ6$ÿA5ÿ C&'3ÿv&ÿ89$ÿUiYchXdrYÿdfhÿfrXifgiÿcfÿUiepiYYcVfÿ8"!#C $''B$DC EF7v2222y22Q9$ÿt%ÿu& 3ÿtu#ÿ"!DÿTÿBB$ÿyvItÿG$ $ÿD3ÿ74'ÿÿ81ÿ74ÿ)9$ÿ@2ÿ&4ÿÿÿ"ÿÿÿ ÿ@ÿ8"!#%%%$D$&$ &&B %$"DBEF$x1xv27$)$$1$1)9$ÿVXfcgd`cVfYÿVpÿ `d`cY`cgdrÿqrcgd`cVfYÿdfhÿfi`eVhY$ÿxÿ819#ÿ1)w 1$ÿ#$x1xD$)$$1$1)ÿ8"!#$&'$x1x(D$)$$1$1)9$ÿyvvtÿ11)x)ÿ8"!#%% %$%&B$&'11)x)9$ ÿ 2 3ÿ2"45ÿ63ÿ7&ÿ89$ÿ@ABÿA'@ÿ8"!#C $''B$DC EFGHIPQ0RS T!'FH29$ÿUVWXY`ÿbcdefVY`cgYÿdfhÿUiepiYYcVfÿqfdrsYcY$ÿt%ÿu& #ÿv!&'&$ÿ!!$ÿwx$ÿyvItÿ1)x)$ C&'&3ÿ6"&ÿ7$5ÿBB3ÿI&ÿ819$ÿ@ÿv@ÿ8"!#C $''B$DC EF1'I%22I2QT !'FH219$ÿ `d`cY`cgdrÿqfdrsYcYÿdfhÿbd`dÿbcYrds$ÿv!&'&ÿvÿTÿIÿ7$ÿ!!$ÿ1w)$ yvItÿ))x)$ & &3ÿ BBDÿv$5ÿ"3ÿ%5ÿ B"3ÿ64ÿ$ÿ819$ÿ@Dÿ&ÿ&4ÿÿÿB%ÿDB@$ÿdfhWVVÿV gVfVi`pcgY$ÿ$ÿB&$ÿ!!$ÿxw$ÿ#$vx)1819xÿ8"!#$&'$(vx)1 (1(x9$ 2'3ÿ&D5ÿ&3ÿ64ÿ$5ÿQ3ÿ&&4ÿ819$ÿ@IH&ÿ6DDÿ&ÿA'ÿy4' ÿB'ÿPB&@ÿ8"!#%%%$&&"'$!&B&D02'!CBx)0IH&0 6DD0&0A'0y4'00B'0PB&B 1x)C1C$!9ÿ8HA9$ pedfcdd`cVfdrÿUiYidpgeÿfi`eVhY$ÿv'$ÿgÿ89#ÿ)w1$ÿ#$)))ÿ8"!#$&'$))( )9$ÿ6&ÿÿADC&ÿx$ 6&ÿ&Dÿ@"!#$% !$&'%h$!"!EBF ()0TBF11@ ijklÿnopqÿrolÿsoltÿquktquÿvwÿxyÿzo{|jÿx}~ÿotÿ}x}y hÿÿBCBÿ&ÿ"ÿ&ÿDDÿ2&Cv"&2B ÿR5ÿBÿ&DÿD4ÿ!!B4$ÿI4ÿ'ÿ" 3ÿ4ÿ'&ÿÿ"ÿ&DÿÿSÿÿH&4ÿHB4$ÿ !ÿÿÿ&'&ÿ&D& ÿÿ"ÿ D 3ÿy$3ÿÿ!&ÿ&'$ "!#$% !$&'% ()0 11.