Download presentation
Presentation is loading. Please wait.
1
5.4 Multiple logistic regression
19
proc logistic data=table4_3 desc; class C/param=glm; model NSa=C W; ods output ParameterEstimates=PE; run; data pe2; set pe; var=compress(Variable||ClassVal0); call symput(var, estimate); data pred; do width=18 to 34 by 0.1; lp=&INTERCEPT+(&C1)+width*&W;C=1; p=exp(lp)/(1+exp(lp));output; lp=&INTERCEPT+(&C2)+width*&W;C=2; p=exp(lp)/(1+exp(lp));output; lp=&INTERCEPT+(&C3)+width*&W;C=3; p=exp(lp)/(1+exp(lp));output; lp=&INTERCEPT+width*&W;; C=4; p=exp(lp)/(1+exp(lp));output; end; proc gplot; plot p*width=C; run;
23
proc logistic data=table4_3 desc; class C/param=glm; model NSa=C W; run; /*test beta for color=0 by comparing the complex model above with simple model below*/ model NSa= W; /*test interaction by comparing the complex model with the model with no interaction*/ model NSa=C W C*W;
27
/. Monotone scoring for color 1,2,3,4
/*Monotone scoring for color 1,2,3,4*/ proc logistic data=table4_3 desc; model NSa=C W; run; /*C1={1,1,1,0} coding for color*/ class C1/param=glm; model NSa=C1 W;
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.