* a);
proc format;
value sesn	1 = 'winter' 
			2 = 'spring'
			3 = 'summer'
			4 = 'fall'
			5 = 'annual';
run;
filename npolar url 'http://www.stat.ncsu.edu/people/rasathurai/courses/st445/npolar.dat';
* Can also read the data as 'it is' and use proc transpose as shown below;
/*
data npolar;
infile npolar firstobs=25 obs=77;
input year 1-4 winter 8-12 spring 15-19 summer 22-26 fall 29-33 annual 36-40;
run;

proc print data = npolar(obs=10);run;

proc transpose data = npolar out=npolar1(rename=(_name_=season col1 = temp));
by year;
run;
proc print data = npolar1(obs=10);run;
*/

data npolar;
infile npolar firstobs=25 obs=77;
input year @;
	do season = 1 to 5;
		input surf @;
		if season = 5 then date =.;
		else date=mdy(season*3-2,1,year);
		output;
	end;
run;

proc print data = npolar(obs=10);
format season sesn. date date10.;run;

* b);

pattern1 symbol color=red;
axis1 label=(a=90 f="Arial/Bold"
"Surface Temperature") order=(-2 to 4 by 1);

axis2 label=
(f="Arial/Bold" "Years") order=(1958 to 2010 by 4 ) /*order=(-730 to 18250 by 1461 )*/;
title1 "Temperature Vs Time";

/* use all seasons except annual
   Note the where statement */
proc gplot data=npolar(where=(season ne 5));
   plot surf*year /*date */=season/vaxis=axis1 haxis=axis2;
   title 'Surface Temperature Over Time';
format date date10. season sesn.;
run;

* (c)avg= (0.7375 + 0.6013 + 0.1955 + 0.3355)/4 = 1.8698/4= 0.46745 ~= annual vaerage = 0.7756;
proc means data = npolar mean std maxdec=4;
class season;
var surf;
format season sesn.;
run;

*(d) Winter has the highest variance as shown in (c);

*e) ;
/* Only use annual surf values */
proc reg data=npolar(where=(season=5)); 
model surf = year; title;
run;