12. cox regression deletion diagnostic

12. cox regression deletion diagnostic -...

Info iconThis preview shows pages 1–5. Sign up to view the full content.

View Full Document Right Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview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elta Betas: IDEA: Measure how coefficient estimates of interest change when we include each observation: Influence of the th observation on : de t- bet = ( ) = - (- ) - th observation removed LS REGRESSION: Can write a formula for ( ) for each coefficient in the model and each observation. Easy to calculate. LOGISTIC AND COX REGRESSION: Exact calculation would require iteration to refit the model for each omitted observation com- putationally intensive. So we approximate ( ) with a one-step estimate: the for- mula can be computed without iteration. HPK-CIH ^ I;JL MKJMHO GO4KI !4F5G *+, 562"78 9:++ !D -(A7"8?% UV_ . stcox status failure _d: censor analysis time _t: time Iteration 0: log likelihood = -505.88396 Iteration 1: log likelihood = -485.31167 Iteration 2: log likelihood = -485.07085 Refining estimates: Iteration 0: log likelihood = -485.07085 Cox regression -- Breslow method for ties No. of subjects = 137 Number of obs = 137 No. of failures = 128 Time at risk = 16663 LR chi2(1) = 41.63 Log likelihood = -485.07085 Prob > chi2 = 0.0000------------------------------------------------------------------------------ _t | Haz. Ratio Std. Err. z P>|z| [95% Conf. Interval]-------------+---------------------------------------------------------------- status | .9673035 .0049074 -6.55 0.000 .9577329 .9769698------------------------------------------------------------------------------ HPK-CIH ^ I;JL MKJMHO GO4KI !4F5G *+, 562"78 9:++ !D -(A7"8?% UV . predict dbstatus, dfbeta . summ dbstatus, de DFBETA - status------------------------------------------------------------- Percentiles Smallest 1% -.0006277 -.0013112 5% -.0005331 -.0006277 10% -.0003114 -.0006112 Obs 137 25% -.0001827 -.0005912 Sum of Wgt. 137 50% -.0000603 Mean -1.78e-12 Largest Std. Dev. .0004324 75% .0001264 .0010132 90% .0003197 .0011798 Variance 1.87e-07 95% .0005921 .0019158 Skewness 3.022487 99% .0019158 .002959 Kurtosis 20.80863 HPK-CIH ^ I;JL MKJMHO GO4KI !4F5G *+, 562"78 9:++ !D -(A7"8?% UVV . stcox status, nohr failure _d: censor analysis time _t: time Iteration 0: log likelihood = -505.88396 Iteration 1: log likelihood = -485.31167 Iteration 2: log likelihood = -485.07085 Refining estimates: Iteration 0: log likelihood = -485.07085 Cox regression -- Breslow method for ties...
View Full Document

This note was uploaded on 02/09/2012 for the course STAT 513 taught by Professor Barbaramc.knight during the Spring '11 term at University of Washington.

Page1 / 13

12. cox regression deletion diagnostic -...

This preview shows document pages 1 - 5. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online