compare_supp_algorithms.SAS

/* Read in the raw names data */

data names_raw;
	infile "C:\Users\dbatten\Desktop\Name Data.csv" dlm = ',' dsd truncover lrecl = 1000;
	
	input first_name :$50. last_name :$50. race :$3. number :5.;
run;

/* Aggreate it at every possible level */
 
proc means noprint data = names_raw;
	class first_name last_name race;
	var number;
 
	output out = names(keep = _type_ first_name last_name race number rename = (_type_ = level)) sum = number;
run;

/* Switch the nulls coming out of PROC MEANS to 'ALL' for convenience */
/* Also, create two copoies - one to suppress with the heuristic, one optimally */
 
data names names_supp names_opt;
	set names;
 
	if first_name = ' ' then first_name = 'ALL';
	if last_name = ' ' then last_name = 'ALL';
	if race = ' ' then race = 'ALL';
run;

/* Run both suppression algorithms */

%suppress(names_supp,number,first_name|last_name|race,3);
%opt_suppress(names_opt,number,first_name|last_name|race,3);


/* Calculate the number of cells suppressed */

proc sql;
	select
		sum(case when s.number is null then n.number else 0 end) as heuristic_suppressed,
		sum(case when o.number is null the n.number else 0 end) as optimal_suppressed,
		sum(n.number) as total_N
	from
		names as n
	left join
		names_supp as s
	on
		n.first_name = s.first_name
		and n.last_name = s.last_name
		and n.race = s.race
	left join
		names_opt as o
	on
		n.first_name = o.first_name
		and n.last_name = o.last_name
		and n.race = o.race;
quit;