options ps=54 ls=76 nodate nonumber;

data q4_2002;
infile 'A:\q4_2002.out';
input   city1    $
        city2    $
        ave_fare
        distance
        passengers
        mktldr   $   /* Airline with highest market share  */
        ldrshare 
        ldrfare
        pricelow $   /*  Airline with lowest price         */
        lowshare
        lowfare;

run;

proc sort;
by city1;
run;

data allcity;
infile 'a:\allcity.dat';
input city $ div $ region $;
city1=city;
region1=region;
run;

proc sort;
by city1;
run;

data q4_2002a;
merge q4_2002 allcity;
by city1;
drop city div region;
run;

proc sort;
by city2;
run;

data allcity;
set allcity;
city2=city;
region2=region;
run;

proc sort;
by city2;
run;

data q4_2002b;
merge q4_2002a allcity;
by city2;
drop city div region;
run;

proc sort;
by city1 city2;
run;

data q4_2002b;
set q4_2002b;
if region1=region2;
if city1 ne city2;
ave_fare=ave_fare/100;
distance=distance/100;
west=0;
midwest=0;
south=0;
if region1='W' then west=1;
if region1='MW' then midwest=1;
if region1='S' then south=1;
run;


* proc print;
run;

proc glm;
model ave_fare=distance west midwest south distance*west distance*midwest distance*south;
run;

proc glm;
model ave_fare=distance west midwest south;
run;

proc glm;
model ave_fare=distance;
run;

proc sort;
by region1;
run;

proc glm;
by region1;
model ave_fare=distance;
run;



quit;