/* Run this program to create ncaa with x1-x42 and avg_grad = y */
/* Here are the variables and their descriptions:
           OBSNUM
           ID
           NAME school name
           AVG_GRAD average 6 yr rate  1996 97 98
           T10 BB tourney wins 90-99
           REGIONAL number appear in regional tourney games
           BPOWER basketball power 1990-98
           RECENT proportion of wins in past 5 years
           CHAMPS number of sport championships - total
           FBWINS PERCENT WINS 90-99 football
           BOWLS bowl appearances past 10 yrs
           ATTEND avg BB  home attendance
           ACCEPT Acceptance rate
           ACT25 ACT COMPOSITE 25TH
           ACT75 ACT COMPOSITE 75TH
           AST_SAL Assistant professor salary
           AVGSAL
           BBRANK10 ranked in BB past 10 years
           BBRANK5 ranked in BB past 5 years
           BOARD room and board
           COST estimated annual total cost
           FBRANK10 football ranked in past 10 years
           FBRANK5 ranked in FB last 5 years
           FBSUM football power 1990-98
           FT % enrolled fulltime
           FT_AID % undergrads receiving aid
           FT_GRAD % first-time undergraduates
           FTIME number enrolled full time
           FULL_SAL Full Professor Salary
           GIVE Giving rank
           GRAD96
           GRAD97 Graduation rate
           GRAD98 1999 GRAD - 1991 FRESHMEN
           IG_PCT % receiving inst grants
           L_PCT % receiving loans
           LOAN loan amount $
           OLD % students 25+
           ONCAMPUS % On living campus
           OUTSTATE % Out of state
           PHD % faculty with PHD
           POP population of city where located
           SF_RATIO Student to faculty ratio
           SIZE Total Enrollment
           SPORTS weighted sports index
           TATEACH % courses taught by TAs
           TOP10 % students in top 10% HS
           TUITION in-state tuition
           WHITE % white
*/




data one;
infile 'K:\www\var.select\ncaa.data.orig.txt' lrecl = 1000 pad;
input obsnum id name $ 10-28 avg_grad   t10  regional  bpower
recent  champs  fbwins  bowls      attend    accept     act25     act75  ast_sal
avgsal  bbrank10  bbrank5   board cost  fbrank10  fbrank5  fbsum
ft  ft_aid  ft_grad     ftime  full_sal give  grad96    grad97  grad98  ig_pct  l_pct
loan  old  oncampus  outstate       phd
pop  sf_ratio      size  sports  tateach  top10  tuition     white;


/* bbindex is the index of basketball strength defined in the journal article */
data two; set one;
bbindex=.889*bbrank10 + .947*t10 + .950*regional + .189*recent;
bbindex=(bbindex-6.4360274)/5.6433947;
size=size/1000;
tuition=tuition/1000;
board=board/1000;

data ncaa;
   set two;
   label avg_grad = "avg_grad"
         top10 = "top10"
         act25 = "act25"
         oncampus = "oncampus"
         ft_grad = "ft_grad"
         size = "size"
         tateach = "tateach"
         bbindex = "bbindex"
         tuition = "tuition"
         board = "board"
         attend = "attend"
         full_sal = "full_sal"
         sf_ratio = "sf_ratio"
         white = "white"
         ast_sal = "ast_sal"
         pop = "pop"
         phd = "phd"
         accept = "accept"
         l_pct = "l_pct"
         outstate = "outstate"
;
rename  avg_grad = y
           top10 = x1
           act25 = x2
           oncampus = x3
           ft_grad = x4
           size = x5
           tateach = x6
           bbindex = x7
           tuition = x8
           board = x9
           attend = x10
           full_sal = x11
           sf_ratio = x12
           white = x13
           ast_sal = x14
           pop = x15
           phd = x16
           accept = x17
           l_pct = x18
           outstate = x19
           t10=x20
           regional=x21
           bpower=x22
           recent=x23
           champs=x24
           fbwins=x25
           bowls=x26
           act75=x27
           avgsal=x28
           bbrank10=x29
           bbrank5=x30
           cost=x31
           fbrank10=x32
           fbrank5=x33
           fbsum=x34
           ft=x35
           ft_aid=x36
           ftime=x37
           give=x38
           ig_pct=x39
           loan=x40
           old=x41
           sports=x42
;
run;
proc means data=ncaa;var y x1-x42;
run;
proc contents data=ncaa;
run;