/****************************************************************/
 /*          S A S   P R O G R A M                               */
 /*                                                              */
 /*    NAME: Bill StJohn
 /*   TITLE: batters2.sas
 /* PRODUCT: SAS
 /*  SYSTEM: UNIX
 /*    KEYS:
 /*   PROCS: PRINT mean, univariate, freq, chart                 */
 /*    DATA: batter2.dat                                         */
 /*                                                              */
 /*     REF:                                                     */
 /*    MISC:                                                     */
 /*    DESC: This SAS program maniputlates data from a file listing
 /*           observances of U.S. pro baseball players (non-pitchers */
 /*           who were born in California or Pennsylvania.
 /*           All but one of the PROCs have been commented out; the
 /*           viewers are encouraged to download this file and
 /*           the data file (batters2.dat) and alter the comments
 /*           to run the other procs.
 /*                I got the data from a web site:
 /*
 /*                  http://www.baseball-reference.com/bio/
 /*                                                              */
 /****************************************************************/

options linesize=140 pagesize=200;
data nodate batters2;

filename batters2 './batters2.dat';
infile batters2;

/*var yr_1st;*/
input
   st        $    1-2    /*  player's home state    */
   na        $    3-22   /*  player's name          */
   yr_1st        24-27   /*  career first year      */
   dec1st        24-26   /*  career first decade    */
   yr_last       29-32   /*  career last  year      */
   ab            38-43   /*  career at bats         */
   hr            62-65   /*  career home runs       */
   rbi           66-70   /*  career rbi             */
   bat_avg        83-86; /*  career batting average */
/*
   bp        $  109-130;
*/
term    = (yr_last - yr_1st);         /*  term    = career length in years.     */
hr_rate = hr / ab * 100;              /*  hr_rate = (Home runs per at-bats) * 100 */
if ab < 100 | term < 5 then delete;   /*  Disregard all players with fewer than */
                                           /*  100 ab-bats or careers shorter than */
                                           /*  five years.                   */
if yr_1st < 1950 then post1950='PRE  ''50';  /*  Indicate whether player began       */
else                  post1950='POST ''50';  /*  career before or after 1950.  */
run;

/***********************************************/
/*  NOTE: Several of the following PROCs (procedures) are commented out. */
/*  Feel free to add or remove comments based  */
/*  on which procs you want to test.           */
/***********************************************/

/*
proc print data=batters2;
run;

proc means data=batters2 mean minimum maximum maxdec=3;
class st post1950;
var bat_avg hr_rate;
title1 'COMPARING BATTING AVERAGES & HOME RUNS AMONG';
title2 'PLAYERS BORN IN CALIF. & PENN., SUB-GROUPING';
title3 'BY PRE- OR POST-1950 CAREER START.';
run;

proc univariate data=batters2 plot;
var bat_avg;
id st;
run;

proc freq data=batters2;
tables dec1st                    term;
label  dec1st = 'Rookie Decade'  term   = 'Years played';
run;

proc chart data=batters2;
hbar st / group=post1950;
label bat_avg='batting avg.' st='state';
run;
*/

proc chart data=batters2;
hbar bat_avg / group=st midpoints= 0 to .4 by .025;
label st='state' bat_avg='Batting avg.';
title1 'COMPARISON OF BATTING AVERAGES AMONG';
title2 'PLAYERS BORN IN TWO U.S. STATES:';
title3 'CALIFORNIA AND PENNSYLVANIA';
run;