* SAS program for linear regression exercise # 1 (VER 14.1);

* data import through menu;
PROC IMPORT OUT= WORK.a 
            DATAFILE= "R:\Data_csv\btb_episodes.csv" 
            DBMS=CSV REPLACE;
     GETNAMES=YES;
     DATAROW=2; 
RUN;

data btb;
  set a;
  intvl_ln=log(intvl); * natural log;
run;

*Q1;
proc univariate data=btb;
  var intvl intvl_ln;
  histogram;
run;

*Q2;
proc reg data=btb;
  model intvl_ln=p_rct;
run;

proc reg data=btb;
  model intvl_ln=p_year;
run;

proc reg data=btb;
  model intvl_ln=hdsize;
run;

*Q3;
proc reg data=btb;
  model intvl_ln=p_rct p_year hdsize;
run;

*Q4;
proc reg data=btb;
  model intvl_ln=hdsize;
  plot intvl_ln*hdsize / overlay pred conf;
  symbol1 v=point i=none c=red;
run;

* extraction of predictions and interval bounds;
proc reg data=btb;
  model intvl_ln=hdsize;
  output out=preds p=fit lcl=predlow ucl=predhigh lclm=cilow uclm=cihigh;
run;

*Q5;
proc reg data=btb;
  model intvl_ln=p_rct p_year hdsize;
run;
  model intvl_ln=p_year hdsize;
run;