ODS RTF FILE = 'C:\Documents and Settings\rjchar2\My Documents\CPH931F09\WA2SAS931F09.rtf';

PROC IMPORT DATAFILE = 'C:\Documents and Settings\rjchar2\My Documents\CPH931F09\WA2DataExcel.xls'
     OUT = Cereal  DBMS = EXCEL REPLACE;
     SHEET = WA2Data;
     GETNAMES = YES;
RUN;

DATA CEREAL;
SET CEREAL;
IF CARBO = -1 THEN CARBO = .;
IF SUGARS = -1 THEN SUGARS = .;
IF POTASS = -1 THEN POTASS = .;
RUN;

DATA CEREAL;
SET CEREAL;
CALFAT = 9*FAT;
CALPRO = 4*PROTEIN;
CALCAR = 4*CARBO;
CALSUG = 4*SUGARS;
RUN;

DATA CEREAL;
SET CEREAL;
RATINGALT = RATING;
IF NAME = 'Cheerios' THEN RATINGALT = 507.64999;
RUN;


proc reg data=cereal;
model rating = calfat;
run;

proc robustreg data=cereal; 
model rating = calfat  ; 
run;


proc reg data=cereal;
model ratingalt = calfat;
run;

proc robustreg data=cereal; 
model ratingalt = calfat  ; 
run;


proc sort data=cereal;
by calfat;
run;

ods output OutputStatistics=cerealstats;

proc loess data=cereal;
model rating = calfat;
run;
symbol1 color=black value=dot h=2 pct;  
symbol2 color=black interpol=join value=none width=1;
   proc gplot data=cerealstats;
      plot (depvar pred)*calfat / overlay 
           hminor = 0
           vminor = 0
           vaxis  = axis1
           frame cframe=ligr;           
	   axis1 label = ( r=20 a=50 ) order=(0 to 100  by 5);
	   title 'LOESS estimation';
   run; 



PROC REG DATA = Cereal;
MODEL RATING  =  CALFAT CALPRO CALCAR CALSUG CALORIES SODIUM FIBER POTASS     / VIF CORRB ALPHA = 0.05 ;
RUN;

  title 'Ridge Trace of Cereal Data';
   symbol1 v=A     c=blue;
   symbol2 v=B     c=yellow;
   symbol3 v=C     c=cyan;
   symbol4 v=D     c=green;
   symbol5 v=E     c=orange;
   symbol6 v=F     c=red;
   symbol7 v=G     c=purple;
   symbol8 v=H     c=pink;
   symbol9 v=I     c=white;
   legend2 position=(middle right inside)
           across=3 cborder=black offset=(0,0)
           label=(color=blue position=(top center)
                  'coefficient estimates') cframe=white;

   proc reg data=cereal
            outest=b ridge=0 to 0.02 by 0.001;
      model RATING = CALFAT CALPRO CALCAR CALSUG CALORIES SODIUM FIBER POTASS  /noprint;
      plot / ridgeplot nomodel legend=legend2 nostat
             vref=0 lvref=1 cvref=blue cframe=ligr;
   run;


title '';
PROC REG DATA = cereal  outseb  outest=ridgeresults  ridge=0.02;
MODEL RATING = CALFAT CALPRO CALCAR CALSUG CALORIES SODIUM FIBER POTASS / VIF CORRB ALPHA = 0.05 ;
RUN;

PROC PRINT DATA=ridgeresults;
RUN;


PROC REG DATA = Cereal;
MODEL RATING  =  FAT PROTEIN CARBO SUGARS CALORIES SODIUM FIBER POTASS ;
PLOT R. * P.;
RUN;
 
PROC REG DATA = Cereal;
MODEL RATING  =  FAT PROTEIN CARBO SUGARS CALORIES SODIUM FIBER POTASS  ;
OUTPUT OUT=cerealnew R = RESID P = PRED;
RUN;


DATA cerealnew;
SET cerealnew;
ABSRESID = abs(RESID);
RUN;

PROC REG DATA = cerealnew;
MODEL ABSRESID = PRED;
OUTPUT OUT=cerealnew2 P = AUXPRED;
RUN;


DATA cerealnew2;
SET cerealnew2;
WEIGHTSFORWLS = 1/(AUXPRED*AUXPRED);
RUN;

PROC REG DATA = cerealnew2;
MODEL RATING  =  FAT PROTEIN CARBO SUGARS CALORIES SODIUM FIBER POTASS  ;
WEIGHT WEIGHTSFORWLS;
RUN;


DATA cerealnew;
SET cerealnew;
PRED2 = PRED*PRED;
RUN;


PROC REG DATA = cerealnew;
MODEL ABSRESID = PRED PRED2;
OUTPUT OUT=cerealnew3 P = AUXPRED;
RUN;


DATA cerealnew3;
SET cerealnew3;
WEIGHTSFORWLS = 1/(AUXPRED*AUXPRED);
RUN;

PROC REG DATA = cerealnew3;
MODEL RATING  =  FAT PROTEIN CARBO SUGARS CALORIES SODIUM FIBER POTASS  ;
WEIGHT WEIGHTSFORWLS;
RUN;

PROC MI DATA=cereal SEED=12345 nimpute=5 out=imputed;
	mcmc chain=multiple displayinit initial=em(itprint);
var RATING FAT PROTEIN CARBO SUGARS CALORIES SODIUM FIBER POTASS;
RUN;
 
PROC PRINT DATA=imputed;
var NAME CARBO SUGARS POTASS;
RUN;
  
PROC REG DATA=imputed outest=outreg covout;
MODEL  RATING  =  FAT PROTEIN CARBO SUGARS CALORIES SODIUM FIBER POTASS;
BY _IMPUTATION_;
RUN;
 
PROC MIANALYZE DATA=OUTREG;
VAR INTERCEPT FAT PROTEIN CARBO SUGARS CALORIES SODIUM FIBER POTASS ;
RUN;

ODS RTF CLOSE;
RUN;