option nodate;
title "Waste percentage";
proc format;
value pl 1="Plant 1" 2="Plant 2" 3="Plant 3" 4="Plant 4" 5="Plant 5";
Data waste;
infile "c:\csc423\week10\waste.txt";
input plant  waste;
format plant pl.;
* define dummy variable if regression model is used;
pl1=(plant=1); pl2=(plant=2);
pl3=(plant=3); pl4=(plant=4);

proc sort;
by plant;

proc means n mean stddev median q1 q3 min max;
by date;
var gmret;
proc boxplot;
plot waste*plant;

proc gplot;
plot waste*plant;
/*proc glm;
class plant;
model waste=plant;
means plant/lsd lines;
run; quit;*/
proc reg;
model waste= pl2 pl3 pl4 pl5;
plot student.*predicted.;
plot npp.*student.;
output student=resid;
run; quit; 
proc univariate;
var resid;
probplot / normal(mu=est sigma=est);
run;