*Model Selection;

*Iris Data;
PROC PRINT data=sashelp.iris;

PROC GLM data=sashelp.iris;
	model PetalWidth=SepalLength;
PROC GLM data=sashelp.iris;
	model PetalWidth=SepalWidth;
PROC GLM data=sashelp.iris;
	model PetalWidth=SepalLength*SepalWidth;
PROC GLM data=sashelp.iris;
	model PetalWidth=SepalLength SepalWidth;
PROC GLM data=sashelp.iris;
	model PetalWidth=SepalLength SepalLength*SepalWidth;
PROC GLM data=sashelp.iris;
	model PetalWidth=SepalWidth SepalLength*SepalWidth;	
PROC GLM data=sashelp.iris;
	model PetalWidth=SepalLength SepalWidth SepalLength*SepalWidth;	
	
PROC GLMSELECT data=sashelp.iris plots=(Criteria Candidates);
	model PetalWidth=SepalLength|SepalWidth/
		selection=stepwise(select=AICC);
		
*Insular Data;		
PROC IMPORT datafile='/home/markwilliamson20/my_courses/markwilliamson0/MW_Datasets_2020/insular.csv'
	dbms=csv out=insular replace; getnames=yes;
PROC PRINT data=insular;

PROC GLMSELECT data=insular plots=(Criteria Candidates);
	model SpeciesRichness=endemic percentendemic Area	altitude elevation
		distanceParamo distancevegisland distancesouthisland distancelargeisland/
		selection=forward(select=AICC);
		
PROC GLMSELECT data=insular plots=(Criteria Candidates);
	model SpeciesRichness=endemic percentendemic Area	altitude elevation
		distanceParamo distancevegisland distancesouthisland distancelargeisland/
		selection=backward(select=AICC);

PROC GLMSELECT data=insular plots=(Criteria Candidates);
	model SpeciesRichness=endemic percentendemic Area	altitude elevation
		distanceParamo distancevegisland distancesouthisland distancelargeisland/
		selection=stepwise(select=AICC);

PROC GLMSELECT data=insular plots=(Criteria Candidates);
	model SpeciesRichness=endemic percentendemic Area	altitude elevation
		distanceParamo distancevegisland distancesouthisland distancelargeisland/
		selection=stepwise(select=AICC);		
*********************************************************************************************************************************;

			
*Scatterplot Matrix;
PROC SGSCATTER data=sashelp.iris;
  matrix SepalLength PetalLength SepalWidth PetalWidth
         / group=species;

PROC SGSCATTER data=insular;
  matrix SpeciesRichness endemic Area altitude elevation distanceParamo
  		/diagonal=(histogram);
*********************************************************************************************************************************;


*Multiple bar/boxplot;
PROC PRINT data=sashelp.cars;

PROC SGPLOT data=sashelp.cars;
	vbox MSRP;
PROC SGPLOT data=sashelp.cars;
	vbox MSRP /category=Drivetrain;
PROC SGPLOT data=sashelp.cars;
	vbox MSRP /category=Drivetrain group=Origin;
	
*Scatter plots using SGPANEL;
PROC SGPANEL data=sashelp.cars;
 panelby Type;
 vbox MSRP /category=Drivetrain group=Origin;

PROC SGPANEL data=sashelp.cars;
 panelby Type Cylinders;
 vbox MSRP /category=Drivetrain group=Origin;
*********************************************************************************************************************************;


*Scatter plots with categories;
DATA cars2; set sashelp.cars;
	lnMSRP=log(MSRP);

PROC SGPLOT data=cars2;
	scatter x=MPG_Highway y=lnMSRP/ markerattrs=(symbol=CircleFilled);
PROC SGPLOT data=cars2;
	scatter x=MPG_Highway y=lnMSRP/ group=Origin markerattrs=(symbol=CircleFilled);