*DATA MINING SAS NOTES
*
*Linear regression and model diagnostics;


*Importing data;

data math5305lab2;
    infile 'C:\Users\jcrawford\Desktop\math5305Lab2Data.txt' dlm=',';
    input Y X1 X2 X3;
proc print data=math5305lab2;
run;



*Multiple regression;
proc reg data=math5305lab2;
    model Y=X1 X2 X3;
run;




*Multiple regression:  Storing residuals in variable called myresiduals;
proc reg data=math5305lab2;
    model Y=X1 X2 X3;
	output out=myoutput
	r=myresiduals;
run;

proc print data=myoutput;
run;

*Testing normality of residuals;
proc univariate data=myoutput normal;
var myresiduals;
run;

*Testing homoscedasticity of residuals (constancy of error variance);
proc reg data=math5305lab2;
    model Y=X1 X2 X3/SPEC;
run;

*Histogram and qqplot for residuals;
proc univariate data=myoutput;
histogram myresiduals/normal;
qqplot myresiduals;
run;


*Plot of Y vs each of the independent variables, X1, X2, and X3;
proc plot data=myoutput;
plot Y*X1 Y*X2 Y*X3;
run;

*Plot of residuals vs each independent variable;
proc plot data=myoutput;
plot myresiduals*X1 myresiduals*X2 myresiduals*X3;
run;


*Plotting within the regression command;
proc reg data=math5305lab2;
    model Y=X1 X2 X3;
	output out=myoutput
	r=myresiduals;
	plot Y*X1 Y*X2 Y*X3;
run;


*Plotting with built in residuals and predicted values (yhat values);
proc reg data=math5305lab2;
    model Y=X1 X2 X3;
	output out=myoutput
	r=myresiduals;
	plot Y*X1 
		 Y*X2 
		 Y*X3
		 residual.*X1 
		 residual.*X2 
		 residual.*X3 
		 Y*predicted. 
		 residual.*predicted.;
run;