options nonotes nosource;
cas mySession sessopts=(CASlib=casuser timeout=100000 locale="en_US");
run;
caslib _all_ assign;

quit;
proc sql;
%if %sysfunc(exist(V0096C01.'DATA'n)) %then %do;
    drop table V0096C01.'DATA'n;
%end;
%if %sysfunc(exist(V0096C01.'DATA'n,VIEW)) %then %do;
    drop view V0096C01.'DATA'n;
%end;
quit;

options validvarname=any;
FILENAME REFFILE DISK '/sasdata/dev/VEA0096/Data/SASLib/SHEI/DATA_SCIENCE/Data_withtheme2.xlsx' lrecl=1000000;

PROC IMPORT DATAFILE=REFFILE 
	DBMS=xlsx
	OUT=V0096C01.'DATA'n;
	GETNAMES=YES;
RUN;
proc sql;
%if %sysfunc(exist(V0096C01.termnum)) %then %do;
    drop table V0096C01.termnum;
%end;
%if %sysfunc(exist(V0096C01.termnum,VIEW)) %then %do;
    drop view V0096C01.termnum;
%end;
quit;



FILENAME REFFILE DISK '/sasdata/dev/VEA0096/Data/SASLib/SHEI/DATA_SCIENCE/termnum.csv';

PROC IMPORT DATAFILE=REFFILE
	DBMS=CSV
	OUT=V0096C01.termnum;
	GETNAMES=YES;
RUN;

PROC CONTENTS DATA=V0096C01.termnum; RUN;
proc sql;
	select max(_TERMNUM_) INTO:term
	from V0096C01.termnum;
quit;


proc sql;
%if %sysfunc(exist(V0096C01.'DATA1'n)) %then %do;
    drop table V0096C01.'DATA1'n;
%end;
%if %sysfunc(exist(V0096C01.'DATA1'n,VIEW)) %then %do;
    drop view V0096C01.'DATA1'n;
%end;
quit;



FILENAME REFFILE DISK '/sasdata/dev/VEA0096/Data/SASLib/SHEI/DATA_SCIENCE/Data_withtheme2.csv' lrecl=1000000;

PROC IMPORT DATAFILE=REFFILE
	DBMS=CSV
	OUT=V0096C01.'DATA1'n;
	GETNAMES=YES;
RUN;

PROC CONTENTS DATA=V0096C01.'DATA1'n; RUN;



proc sql;
%if %sysfunc(exist(V0096C01.'DATA2'n)) %then %do;
    drop table V0096C01.'DATA2'n;
%end;
%if %sysfunc(exist(V0096C01.'DATA2'n,VIEW)) %then %do;
    drop view V0096C01.'DATA2'n;
%end;
quit;



FILENAME REFFILE DISK '/sasdata/dev/VEA0096/Data/SASLib/SHEI/DATA_SCIENCE/Data_withtheme2.csv' lrecl=1000000;

PROC IMPORT DATAFILE=REFFILE
	DBMS=CSV
	OUT=V0096C01.'DATA2'n;
	GETNAMES=YES;
RUN;

PROC CONTENTS DATA=V0096C01.'DATA2'n; RUN;

proc sql;
%if %sysfunc(exist(V0096C01.'DATA3'n)) %then %do;
    drop table V0096C01.'DATA3'n;
%end;
%if %sysfunc(exist(V0096C01.'DATA3'n,VIEW)) %then %do;
    drop view V0096C01.'DATA3'n;
%end;
quit;



FILENAME REFFILE DISK '/sasdata/dev/VEA0096/Data/SASLib/SHEI/DATA_SCIENCE/Data-withtheme2.csv' lrecl=1000000;

PROC IMPORT DATAFILE=REFFILE
	DBMS=CSV
	OUT=V0096C01.'DATA3'n;
	GETNAMES=YES;
RUN;

PROC CONTENTS DATA=V0096C01.'DATA3'n; RUN;

ods noproctitle;
%macro create(howmany);
 %LET x=1;
 %LET term=&term;
 %LET indeed=%sysfunc(open(V0096C01.DATA));
 %DO %WHILE (%sysfunc(varnum(&indeed,cconverted&x.))>0);



  %LET y=%sysevalf(&x-1);
  %put &x;
  %IF &x>1 %THEN %DO;
   data _null_;
    if 0 then set V0096C01.modified&y. nobs=n;
    put n;
    stop;
   run;
   
   
  %END; 

  %IF &x=1 %then %do;
   proc gradboost data=V0096C01.DATA maxdepth=50 outmodel=V0096C01.DATA2;
	partition fraction(validate=0.12 test=0.12);
	target cconverted&x. / level=nominal;
	input hello1-hello&term./ level=interval;
	input 'Feedback Type'n 'Feedback Channel'n 'Digital Product'n 'Service Type'n 'Service Detail'n 'Reason Level 1'n 'Feedback Provider Type'n 'Responsible group'n 'Reason Level 2'n/ level=nominal;
	autotune tuningparameters=(ntrees(init=150) samplingrate vars_to_try(UB=1000 init=300) 
		learningrate lasso ridge) objective=misc;
	ods output TunerResults=_tempTuneResults_;
	ods output EvaluationHistory=_tempEvalHistory_;
	ods output IterationHistory=_tempIterHistory_;
	ods output FitStatistics=Work._Gradboost_FitStats_;
	score out=V0096C01.DATA1 copyvars=(cconverted&x. Index );
	savestate rstore=V0096C01.DATA3;
	id Index;
   run;
  %end; 
  %else %do;
   
   proc gradboost data=V0096C01.modified&y. maxdepth=50 outmodel=V0096C01.DATA2;
	partition fraction(validate=0.12 test=0.12);
	target cconverted&x. / level=nominal;
	input  hello1-hello&term./ level=interval;
	input I_cconverted1-I_cconverted&y. 'Feedback Type'n 'Feedback Channel'n 'Digital Product'n 'Service Type'n 'Service Detail'n 'Reason Level 1'n 'Feedback Provider Type'n 'Responsible group'n 'Reason Level 2'n/ level=nominal;
	autotune tuningparameters=(ntrees(init=150) samplingrate vars_to_try(UB=1000 init=300) 
		learningrate lasso ridge) objective=misc;
	ods output TunerResults=_tempTuneResults_;
	ods output EvaluationHistory=_tempEvalHistory_;
	ods output IterationHistory=_tempIterHistory_;
	ods output FitStatistics=Work._Gradboost_FitStats_;
	score out=V0096C01.DATA1 copyvars=(cconverted&x. Index );
	savestate rstore=V0096C01.DATA3;
	id Index;
   run;
  %end;

  proc sgplot data=Work._Gradboost_FitStats_;
	title3 'Misclassifications by Number of Trees';
	title4 'Training vs. Validation';
	series x=Trees y=MiscTrain;
	series x=Trees y=MiscValid /lineattrs=(pattern=dot thickness=2);
	yaxis label='Misclassification Rate';
	label Trees='Number of Trees';
	label MiscTrain='Training';
	label MiscValid='Validation';
  run;

  title3;
  title 'Tune with Genetic algorithm';

  data _null_;
	set _tempTuneResults_;

	if _n_=1 then
		call symput('default', misclasserr);

	if _n_=2 then
		call symput('best', misclasserr);
  run;

  data _tempEvalHistory_;
	set _tempEvalHistory_;
	default=&default;
	best=&best;
  run;

  proc sgplot data=_tempEvalHistory_;
	scatter x=evaluation y=misclasserr / group=iteration;
	series x=evaluation y=default/curvelabel="default" 
		lineattrs=(pattern=shortdash);
	series x=evaluation y=best / curvelabel="best" lineattrs=(pattern=shortdash 
		color=green);
  run;

  proc sgplot data=_tempIterHistory_;
	series x=iteration y=best_obj;
	series x=iteration y=time_sec / y2axis;
  run;

  proc delete data=_tempTuneResults_ _tempEvalHistory_ _tempIterHistory_;
  run;

  proc delete data=Work._Gradboost_FitStats_;
  run;
  proc astore;
   download rstore=V0096C01.DATA3
            store="/sasdata/dev/VEA0096/Data/SASLib/SHEI/DATA_SCIENCE/svm1store&x.sasast";
  quit;
  %IF &x=1 %then %do;
   proc SQL;
    create table modified&x. as select * from V0096C01.DATA a INNER JOIN V0096C01.DATA1 b ON a.index=b.index; 
   QUIT;
   DATA V0096C01.modified&x.;
    Set modified&x.;
   Run;
  %end;
  %ELSE %DO;
   proc SQL;
    create table modified&x. as select * from V0096C01.modified&y. a INNER JOIN V0096C01.DATA1 b ON a.index=b.index; 
   QUIT;
   DATA V0096C01.modified&x.;
    Set modified&x.;
   Run;
  %END;
  proc sql;
    %if %sysfunc(exist(V0096C01.'DATA1'n)) %then %do;
     drop table V0096C01.'DATA1'n;
    %end;
    %if %sysfunc(exist(V0096C01.'DATA1'n,VIEW)) %then %do;
     drop view V0096C01.'DATA1'n;
    %end;
  quit;



    FILENAME REFFILE DISK '/sasdata/dev/VEA0096/Data/SASLib/SHEI/DATA_SCIENCE/Data_withtheme2.csv' lrecl=1000000;

    PROC IMPORT DATAFILE=REFFILE
	DBMS=CSV
	OUT=V0096C01.'DATA1'n;
	GETNAMES=YES;
    RUN;
  %LET x=%sysevalf(&x+1);
 %end;
 %Let y=%sysevalf(&x-1);
 DATA targetnumber;
  number=&y;
 RUN;
%mend create;
%create(10);

filename xout DISK '/sasdata/dev/VEA0096/Data/SASLib/SHEI/DATA_SCIENCE/targetnumber.csv' lrecl=1000000;
proc export file=xout data=targetnumber dbms=xlsx  replace;
run;