libname stat541 '/home/grego1/STAT 541/';
run;

*Placing LENGTH in the right place;

data ecoli;
set stat541.ecoli;
*This actually generates a warning;
length station $9.;
run;

proc contents data=ecoli;
run;

data ecoli;
length station $9.;
set stat541.ecoli;
run;

proc contents data=ecoli;
run;

*Truncation and PROC COMPARE;
data ecolirt;
length ratio 4.;
set stat541.ecoli;
ratio=ecoli/fecalcoli;
run;

data ecolir;
set stat541.ecoli;
ratio=ecoli/fecalcoli;
run;

*There are 135 cases with no differences--let's look at them;
*Note that record 36 is missing, for instance;
proc compare base=ecolir compare=ecolirt out=ecolicomp outdif maxprint=100;
run;

data ecolicomp;	set ecolicomp (rename=(_obs_=obsno));
aratio=abs(ratio);
drop ratio;
run;

title 'Ecoli/Fecal Coliform';
*Let's print observations with an absolute difference of 0;
proc sql outobs=150;
select r.record, r.ecoli, r.fecalcoli, r.ratio, c.aratio from ecolir as r inner join ecolicomp as c on
r.record=c.obsno
order by aratio;
quit;

*Some of these look as though diff should be 0;
proc sql outobs=50;
select r.record, r.ecoli, r.fecalcoli, r.ratio, c.aratio from ecolir as r inner join ecolicomp as c on
r.record=c.obsno
order by aratio descending;
quit;
title;

*Compress data sets with different options.  Fall2008 and fall2009 are;
*permanent SAS data sets;

libname stat541 '/home/grego1/STAT 541';
run;

data f08comp (compress=yes); set stat541.fall2008;
data f09comp (compress=binary); set stat541.fall2009;
run;

*Create a random sample with a lot of repeats;

proc format;
value tffmt 1="T"
            0="F";
			run;
data tftest (drop=i);
array quiz $ q1-q40;
do i=1 to 100000;
do over quiz; 
  quiz=put(rand("BERNOULLI",0.5),tffmt.);
*  quiz=put(ceil(2*ranuni(0)-1),tffmt.);
end;
output;
end;
run;

data tfcomp (compress=binary); set tftest;
run;