/* OpenIntro LAB 7   */

*Read in the data from the CSV file located on the OpenIntro website;
filename mlb11 url 
         'http://www.openintro.org/stat/data/mlb11.csv';

proc import datafile=mlb11
            out=mlb11
            dbms=csv
            replace; 
   getnames=yes;
run;

*Calculate correlation between runs and at_bats;
proc corr data=mlb11;
   var runs at_bats;
run;

*Plot regression line for runs and at_bats;
proc sgplot data=mlb11;
   reg x=at_bats y=runs;
run;

*Estimate the intercept and slope of the regression line;
proc reg data=mlb11;
   model runs=at_bats;
run; 
quit;

*Plot regression line for runs and at_bats;
proc sgplot data=mlb11;
   reg x=at_bats y=runs;
run;

*Re-run regression with each graph output separately;
proc reg data=mlb11 plots=diagnostics(unpack);
   model runs=at_bats;
run;
quit;