*********************************************************************	
	
SUPPLEMENTAL FILE 3 - Create fake dataset sampled from real HDSS data	
	
Author: Christopher Rentsch	
Email: Christopher.Rentsch@lshtm.ac.uk	
	
*********************************************************************;	
	
	
*set up library;	
options nocenter ps=500 ls=250;	
"libname REAL ""\\filepath"";"	
"libname FAKE ""\\filepath"";"	
	
	
*bring in real HDSS dataset;	
data real;	
set REAL.HDSSDATA;	
run;	
proc contents data = real;	
run;	
	
	
*create ID;	
data ID;	
length ID 3.; ID = 1; do ID=1 to 100; output; end; format ID 3.;	
run;	
proc print data = ID noobs;	
run;	
	
	
"*separately and for each field, take a random sample of 100 values from real HDSS data;"	
*and create an ID 1 to 100;	
*field1;	
data one;	
set real (keep=ID field1);	
rand=rand('UNIFORM');	
length ID 3.; ID = 1; do ID=1 to 100; output; end; format ID 3.;	
proc sort data = one;	
by rand;	
data field1;	
set one (firstobs=1 obs=100);	
run;	
proc print data = field1;	
run;	
	
	
	*example -- FirstName;
	*FirstName;
	data one;
	set real (keep=FirstName);
	rand=rand('UNIFORM');
	proc sort data = one;
	by rand;
	data FirstName;
	set one (firstobs=1 obs=100);
	run;
	proc print data = FirstName;
	run;
	
	
	
*bring all fields together;	
data FAKE.fake100;	
merge field1-field25;	
by ID;	
run;	
	
	
*NB. some manual edits were performed within excel on the fake data to allow 	
the case examples to better mimic reality in the field;