/* 2018.03.12.
   6. gyakorlat - SAS
   GRAPH, Grafikus megjelenitesek SAS Studio-val
*/

/* SASHELP.CLASS adatallomany megismerese */
proc print data=sashelp.class;
run;

/* Sajat hasonlo adatok letrehozasa */
data xclass;
 input nem $ 1. kor korcsop mag kg;
 cards;
1 15 2 154 58
2 16 3 154 53
1 15 2 160 60
1 15 2 153 57
1 14 2 159 58
2 15 2 150 49
2 16 3 162 60
1 16 3 168 64
2 17 3 170 59
2 13 1 149 53
2 14 2 152 53
1 15 2 170 70
1 14 2 174 72
1 13 1 168 67
2 12 1 146 46
1 12 1 145 47
2 12 1 150 50
2 12 1 149 50
1 13 1 151 54
2 15 2 162 56
1 16 3 177 74
2 14 2 160 60
;
run;

/* Hasonlo, mint a Task\Data\Transform data, lehet egyenileg definialni - nem pont igy */
data xclass;
 set xclass;
 if nem = 1 then nem_f = "Ferfi";
 else nem_f = "No";
 if korcsop = 1 then kcs = "12-13";
 else if korcsop = 2 then kcs = "14-15";
 else kcs = "16";
run;

/********** Leiro statisztika *************/
/* Tasks\Statistics\Data exploration - 
   Classification variables: diszkret valtozok, mozaic plot */
options validvarname=any;
ods noproctitle;
ods graphics / imagemap=off;
proc freq data=WORK.XCLASS;
	ods select MosaicPlot;
	tables nem_f*kcs / plots=mosaicplot(square);
run;
ods graphics / imagemap=on;


/* Continuous variables: kor, mag, kg */
options validvarname=any;
ods noproctitle;
ods graphics / imagemap=on;
/* Scatter plot matrix macro */
%macro scatterPlotMatrix(xVars=, title=, groupVar=);
	proc sgscatter data=WORK.XCLASS;
		matrix &xVars / %if(&groupVar ne %str()) %then
			%do;
				group=&groupVar legend=(sortorder=ascending) %end;
		diagonal=(histogram kernel normal);
		title &title;
	run;
	title;
%mend scatterPlotMatrix;
/* Histogram and box plot template */
proc template;
	define statgraph histobox;
		dynamic AVAR;
		begingraph;
		entrytitle "Distribution of " eval(catq('q', colname(AVAR)));
		layout lattice / rows=2 columndatarange=union rowgutter=0 rowweights=(0.75 
			0.25);
		layout overlay / yaxisopts=(offsetmax=0.1) xaxisopts=(display=none);
		layout gridded / columns=2 border=on autoalign=(topright topleft);
		%let _lft = halign=left;
		%let _rgt = halign=right;
		entry &_lft "Mean";
		entry &_rgt eval(strip(put(mean(AVAR), best.)));
		entry &_lft "Std Dev";
		entry &_rgt eval(strip(put(stddev(AVAR), best.)));
		entry &_lft "N";
		entry &_rgt eval(strip(put(n(AVAR), best.)));
		endlayout;
		histogram AVAR /;
		endlayout;
		layout overlay /;
		BoxPlot Y=AVAR / orient=horizontal;
		endlayout;
		endlayout;
		endgraph;
	end;
run;

%scatterPlotMatrix(xVars=kor mag kg, title="Scatter plot matrix", groupVar=);

proc sgrender data=WORK.XCLASS template=histobox;
	dynamic AVAR='kor';
run;
proc sgrender data=WORK.XCLASS template=histobox;
	dynamic AVAR='mag';
run;
proc sgrender data=WORK.XCLASS template=histobox;
	dynamic AVAR='kg';
run;


/* Csoportonkenti eloszlasok: kg ~ kcs 
   Continuous variable: kg
   Classification variable: kcs
*/
options validvarname=any;
ods noproctitle;
ods graphics / imagemap=on;
/* Histogram (one-way or two-way) */
%macro DEHisto(data=, avar=, classVar=);
	%local i numAVars numCVars cVar cVar1 cVar2;
	%let numAVars=%Sysfunc(countw(%str(&avar), %str( ), %str(q)));
	%let numCVars=%Sysfunc(countw(%str(&classVar), %str( ), %str(q)));
	%if(&numAVars>0 & &numCVars>0) %then
		%do;
			%if(&numCVars=1) %then
				%do;
					%let cVar=%scan(%str(&classVar), 1, %str( ), %str(q));
					proc sql noprint;
						select count(distinct &cVar) into :nrows from &data;
					quit;
					/* One-way histogram */
					proc univariate data=&data noprint;
						var &avar;
						class &cVar;
						histogram &avar / nrows=&nrows
             normal(noprint) kernel;
						inset mean std n / position=ne;
					run;
				%end;
			%else
				%do;
					/* One-way histogram of each class variable */
					%do i=1 %to %eval(&numCVars);
						%let cVar=%scan(%str(&classVar), &i, %str( ), %str(q));
						proc sql noprint;
							select count(distinct &cVar) into :nrows from &data;
						quit;
						proc univariate data=&data noprint;
							var &avar;
							class &cVar;
							histogram &avar / nrows=&nrows
                 normal(noprint) kernel;
							inset mean std n / position=ne;
						run;
					%end;
					/* Two-way histogram */
                %let cVar1=%scan(%str(&classVar), 1, %str( ), 
						%str(q));
					%let cVar2=%scan(%str(&classVar), 2, %str( ), %str(q));
					proc sql noprint;
						select count(distinct &cVar1) into :nrows from &data;
					quit;
					proc sql noprint;
						select count(distinct &cVar2) into :ncols from &data;
					quit;
					proc univariate data=&data noprint;
						var &avar;
						class &cVar1 &cVar2;
						histogram &avar / nrows=&nrows ncols=&ncols
             normal(noprint) kernel;
						inset mean std n / position=ne;
					run;
				%end;
		%end;
%mend DEHisto;
%DEHisto(data=WORK.XCLASS, avar=kg, classVar=kcs);

proc sort data=WORK.XCLASS out=WORK.TempSorted4877;
	by kcs;
run;
proc boxplot data=WORK.TempSorted4877;
	plot (kg)*kcs / boxstyle=schematic;
run;
proc datasets library=WORK noprint;
	delete TempSorted4877;
run;


/********* Summary statistics *********/
/*  Analysis variable: mag
 *  Classification variable: kcs
 */
ods noproctitle;
ods graphics / imagemap=on;
proc means data=WORK.XCLASS chartype mean std min max median n nmiss vardef=df 
		skewness kurtosis p5 q1 q3 p95 qmethod=os;
	var mag;
	class kcs;
run;
proc univariate data=WORK.XCLASS vardef=df noprint;
	var mag;
	class kcs;
	histogram mag / normal(noprint) kernel;
	inset mean std min max median n nmiss skewness kurtosis p5 q1 q3 p95 / 
		position=ne;
run;
proc sort data=WORK.XCLASS out=WORK.TempSorted2236;
	by kcs;
run;
proc boxplot data=WORK.TempSorted2236;
	plot (mag)*kcs / boxstyle=schematic;
	inset mean stddev min max nobs / position=ne;
	insetgroup mean stddev min max n q1 q2 q3 / position=top;
run;
proc datasets library=WORK noprint;
	delete TempSorted2236;
run;



/*********** Distribution analysis ***********/
/*  Analysis variable: mag
 *  Classification variable: nem_f
 */
ods noproctitle;
ods graphics / imagemap=on;
/*** Exploring Data ***/
proc univariate data=WORK.XCLASS;
	ods select Histogram;
	var mag;
	class nem_f;
	histogram mag / normal kernel;
	inset n / position=ne;
run;
proc univariate data=WORK.XCLASS;
	ods select Histogram GoodnessOfFit QQPlot;
	var mag;
	/*** Checking for Normality ***/
	histogram mag / normal(mu=est sigma=est);
	inset mean std skewness kurtosis n / position=ne;
	qqplot mag / normal(mu=est sigma=est);
	inset mean std skewness kurtosis n / position=nw;
run;



/***************** Tasks\Graph\Bar Chart *****************/
/*  Category: nem_f
 *  Subcategory: kcs
 */
ods graphics / reset width=6.4in height=4.8in imagemap;
proc sgplot data=WORK.XCLASS;
	title height=14pt "Gyakorisági oszlopdiagram";
	footnote2 justify=left height=12pt "Szeged, 2018. március 12.";
	vbar nem_f / group=kcs groupdisplay=cluster;
	yaxis grid;
run;
ods graphics / reset;
title;
footnote2;


/* Kis modositassal:
   Ne a gyakorisagokat, hanem folytonos valtozo: kg atlag ertekeit mutassuk +/- SE
*/
ods graphics / reset width=6.4in height=4.8in imagemap;
proc sgplot data=WORK.XCLASS;
	title height=14pt "Átlag-szórás diagram (átlag +/- SE)";
	footnote2 justify=left height=12pt "Szeged, 2018. március 12.";
	vbar nem_f / response=kg group=kcs groupdisplay=cluster datalabel limits=both 
		limitstat=stderr stat=mean;
	yaxis grid;
run;
ods graphics / reset;
title;
footnote2;

/* BARWIDTH parameter allitasa 
   + szurkearnyalatos kep beallitasa: ODS style=journal
   es kimentese: gpath, imagename */
/* Google: sas sgplot vbar 
   http://support.sas.com/documentation/cdl/en/grstatproc/62603/HTML/default/viewer.htm#vbar-stmt.htm
   https://support.sas.com/resources/papers/proceedings15/2441-2015.pdf
*/
ODS LISTING GPATH = '/home/laszlo.anna/a2018'; 
ods listing style=journal;
ods graphics / reset width=6.4in height=4.8in imagemap imagename='plot1';
proc sgplot data=WORK.XCLASS;
	title height=14pt "Átlag-szórás diagram (átlag +/- SE)";
	footnote2 justify=left height=12pt "Szeged, 2018. március 12.";
	vbar nem_f / response=kg group=kcs groupdisplay=cluster datalabel limits=both 
		limitstat=stderr stat=mean barwidth=.8 ;
	yaxis grid;
run;
ods graphics / reset;
title;
footnote2;




/*********** Scatter plot **********/
/*  X: mag 
 *  Y: kg
 *  Group: nem_f
 *  esetleges modositas: reg ... / ... degree=1 clm='CL Mean'
 */
ods graphics / reset width=6.4in height=4.8in imagemap;
proc sgplot data=WORK.XCLASS;
	reg x=mag y=kg / nomarkers group=nem_f;
	scatter x=mag y=kg / group=nem_f;
	xaxis grid;
	yaxis grid;
run;
ods graphics / reset;





/************ Snippets\Graph *************/
/*--Bar Panel--*/
title 'Mileage by Origin and Type';
proc sgpanel data=sashelp.cars(where=(type in ('Sedan' 'Sports'))) noautolegend;
  panelby Type / novarname columns=2 onepanel;
  vbar origin / response=mpg_city stat=mean group=origin;
  rowaxis grid;
run;

/*--Fit Plot--*/
title 'Mileage by Horsepower for USA';
proc sgplot data=sashelp.cars(where=(origin='USA'));
  reg x=horsepower y=mpg_city / degree=2 clm='CL Mean';
  keylegend / location=inside position=topright across=1;
  xaxis grid;
  yaxis grid;
run;