daynebatten

/* Sort so we can transpose */

proc sort data = joined;
	by county degree;
run;

/* And transpose! */

proc transpose data = joined out = constraints(drop = _NAME_) prefix = con;
	by county degree;

daynebatten

/*	target_data is the data set you want to suppress
	dimensions is a pipe-separated list of dimensions to suppress */
 
%macro constraints(target_data, dimensions);
	/* Count the number of dimensions we're working with */
	 
	%let num_dims = %eval(%sysfunc(count(&dimensions, |)) + 1);
 
	/* Build a macro variable array with the dimension names */
 

daynebatten

/* Minimize sum of cells suppressed, with a small punishment for suppressing an unnecssary number of cells (e.g., suppressing all 0's) */
	
minimize target = (sum{i in dec_set} values[i] * choices[i]) + (sum{i in dec_set} choices[i] * .0001);

daynebatten

/* Primary suppressed cells must be suppressed in solution */

con primary_con {i in dec_set}: choices[i] >= primary[i];

daynebatten

/* These decisions variables govern whether we suppress 0 or 2+ cells within each grouping constraint */

var choose_zero {con_set} binary;

/* For each grouping constraint, we must have 0 cells suppressed if we're choosing to have 0 cells suppressed */

con zero_con {i in con_set}: (sum{j in dec_set} choices[j] * constraints[i, j]) - (1 - choose_zero[i]) * 100000000000 <= 0;

/* For each grouping constraint, we must have 2+ cells suppressed if we're choosing to have 2+ cells supppressed */

daynebatten

/*	We need constraints for every combination of dimensions you can build by leaving out one dimension at a time.
	This is pretty simple in a 2-dimensional dataset. */

proc sql;
	create table
		levels

	/* Constraints across the degree dimension */

	as select distinct

daynebatten

/*	target_data is the data set you want to suppress
	target_var is the variable you want to suppress
	dimensions is a pipe-separated list of dimensions to suppress
	cutoff is the highest value you want to suppress */

%macro opt_suppress(target_data, target_var, dimensions, cutoff);
	/* Count the number of dimensions we're working with */
	 
	%let num_dims = %eval(%sysfunc(count(&dimensions, |)) + 1);
 

daynebatten

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;
 
proc means noprint data = names_raw;
	class first_name last_name race;
	var number;
 

daynebatten

/* 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 */
 

daynebatten

   left daughter right daughter split var split point status prediction
1              2              3         3        2.45      1          0
2              0              0         0        0.00     -1          1
3              4              5         4        1.75      1          0
4              6              7         3        5.05      1          0
5              8              9         3        4.85      1          0
6             10             11         4        1.65      1          0