*%%--First dataset--%%;

data brand;
infile 'CH06PR05.DAT';
input Y X1 X2;
run;
proc print;run;

*6.5(a);
proc gplot data=brand;
plot Y*X1 Y*X2 X1*X2;
run;

proc corr data=brand noprob;
var Y X1 X2;
run;

*6.5(b)(c)(d), 6.6(a,b);
proc reg data=brand;
model Y=X1 X2/p r;
output out=brandres p=pred r=res;
run;

proc print data=brandres;run;

proc gplot data=brandres;
plot res*pred res*X1 res*X2/vref=0;
run;

proc univariate data=brandres plot;
var res;
qqplot res;
run;

*6.6(c)-Bonferroni correction t value;
data bonf;
tc=tinv(0.9975,13);
run;
proc print;run;

*6.8;
data b2;
X1=5; 
X2=4;
output;
run;
proc print;run;

data brand2;
set brand b2;
run;
proc print;run;

proc reg data=brand2;
model Y=X1 X2/p clm cli alpha=0.01;
run;

*7.3 7.12;
proc reg data=brand;
model Y=X1 X2/ss1 ss2 pcorr1 pcorr2;
run;

* 7.24;
proc reg data=brand;
model Y=X1;
run;

*7.30;
proc reg data=brand;
model Y X1 = X2 / r;
output out=double r=rY r1;
run;
proc print data=double;run;

proc corr data=double noprob;
var rY r1;
run;

*10.9;
proc reg data=brand;
model Y = X1 X2/ r influence;
run;

*%%--Second dataset--%%;

data value;
infile 'CH08PR24.DAT';
input Y X1 X2;
run;
proc print;run;

proc sort data=value;
by X1 X2;
run;

symbol1 v=1 i=sm70 c=black l=1;
symbol2 v=2 i=sm70 c=black l=3;
proc gplot data=value;
   plot Y*X1=X2;
run;

data value2;
set value;
X12=X1*X2;
run;
proc print;run;

proc reg data=value2;
   model Y = X1 X2 X12;
   sameline: test X2, X12;
run;

*%%--Third dataset--%%;
data job;
infile 'CH09PR10.DAT';
input Y X1-X4;
run;
proc print;run;

*9.10(a) a univariate analysis is important
before doing regression; 
proc univariate data=job plot;
var Y X1-X4;
qqplot Y X1-X4;
run;

*9.10(b,c);
proc corr data=job noprob;run;

proc reg data=job;
model Y=X1-X4;
run;

*9.11;
proc reg data=job;
model Y=X1-X4/selection=adjrsq rsquare cp aic press b;
run;

*9.18;
proc reg data=job;
model Y=X1-X4/selection=forward alpha=0.1;
run;

proc reg data=job;
model Y=X1-X4/selection=forward alpha=0.05;
run;

proc reg data=job;
model Y=X1-X4/selection=backward alpha=0.05;
run;

proc reg data=job;
model Y=X1-X4/selection=stepwise alpha=0.05;
run;