/*      
     Q:  If you had to choose between these artists 
         for the grammy, 
         which would you pick?
            (0) Lady Gaga
            (1) Stevie Wonder
     Q:   How old are you? _____
*/

%let cut 27.2; 

run; 
proc format; 
value winner 1="Wonder" 0="Gaga" 
other = "-";
value old 0="Age < &cut" 1="Age > &cut";
title "Grammy survey";

****************** Create Some Data *****************; 
data grammy;
  do i=1 to 30000;  train = (i<2001); drop i; 
     Age = round(40+8*normal(1827655)); 
     age = max(age,8); 
	 p = exp(.1*(Age-30))/(1+exp(.1*(Age-30))); 
	 Y = (ranuni(123)<p); 
output; end; 
run; 

/****************** Graphics ***********************;  
run; 
data graphics; set grammy; where train=1;
YY=Y+.1*normal(123); * Jitter points so they all show *; 
proc sort; by age p y; 
axis1 offset = (5,5) minor=none major=none;  
proc gplot data=graphics; 
 plot Y*Age  p*age/overlay vaxis=axis1 href=&cut; 
   format  Y winner.; 
 plot YY*Age p*age/overlay vaxis=axis1 href=&cut; 
   format YY winner.; 
   symbol1 v=plus c=black i=none h=.5; 
   symbol2 v=dot c=red i=none; 
proc gchart; vbar age/subgroup=Y; 
proc means n data=graphics noprint; 
  var Y; by age p Y;  
  output out=graph2 n=count; 
data graph2; set graph2; 
  Z=count; color="blue"; shape="balloon";  output; 
  Z=8; Y=p; color="red"; shape = "prism"; output; 
proc g3d;  
  scatter Y*Age=Z /color=color shape=shape zmin=0 rotate=10; 
  label Z="n";  
proc gplot; bubble Y*age=count;

*  */;
run; 
************ end graphics **************************; 
run; 


*****************************************************/; 

data next; set grammy; old=(age>&cut); 
format Y winner. old old.;
proc freq data=next; tables old*Y/ nocol nopercent chisq; where train=1; 
title "Cut point at &cut";
output out=out1 chisq;
data out1; 
set out1; cut=&cut; 
  if p_pchi>0 then logworth=-1*log10(p_pchi) - log10(49);
else logworth=.;  *** There are 50 age groups -> 49 splits possible **; 
keep cut logworth _pchi_ p_pchi; 
proc append base=results data=out1;
proc sort; by cut;
proc print data=results; 
run;
proc plot data=results; 
plot logworth*cut/vpos=20 hpos=40; 
title "Logworth for Cutoffs Thus Far"; 
run; 

/* 
Proc datasets; Delete results; run;     

*/