libname lib "C:\Users\anton.bekkerman\Desktop\SAS" ; 

data lib.dt2 ; 
set lib.dt1 ;
/* Remove observations with missing values */
if year <= 1999 then delete ;
/* Keep only certain variables */
keep chickens acres ;
run ;

data lib.dt1;
set lib.dt1;
/* Create dummy variable for indicating year */
if year < 2000 then early=1 ; 
else early = 0 ;

/* Create constant */
constant = 5;

/* Index corresponding to observation number */
index = _n_ ; 

run;

/* Sort in ascending order by chickens */
proc sort data=lib.dt1 ;
by chickens ;
run;
/* Create index based on # of chickens */
data lib.dt1; 
set lib.dt1;
index = _n_;
run;

/* Merging data sets by using an index */
/* (1) Sort both data sets by the index on which merging occurs */
proc sort data=lib.planted;
by fips;
run;
proc sort data=lib.yield;
by fips;
run;

/* (2) Merge using data step */
data lib.merged;
merge lib.planted lib.yield ;
by fips ;
run;


/* Day 2 */

/* Merge planted and yield data by fips and by year */

proc sort data=lib.planted;
by fips year;
run;
proc sort data=lib.yield;
by fips year;
run;

data lib.merged;
merge lib.planted lib.yield;
by fips year;
run;

/* Summary */
proc means data=lib.merged;
run;

/* Summary stats by year */
proc sort data=lib.merged;
by year;
run;
proc means data=lib.merged;
by year;
var planted harvested yield;
run;

/* Summary stats, overall -- mean, std, median, observations, missing values */
proc means data=lib.merged mean std median n nmiss;
var planted harvested yield;
output out=lib.summ mean=/autoname;
run;

/* Distributional properties */
proc freq data=lib.merged;
table yield;
run;

/* Summary of the distributional properties */
proc univariate data=lib.merged;
/* Only look at years between 2001 and 2005 */
where 2001<=year<=2005;
var yield;
run;

/* Graphing */
proc sgplot data=lib.merged;
where fips = 30001;
xaxis type=discrete;
series x=year y=harvested / lineattrs=(color=green thickness=4);
series x=year y=planted;
scatter x=year y=harvested;
run;

proc sgplot data=lib.merged;
histogram yield;
density yield / type=kernel;
run;






/* Matrix language */
proc iml; 

a = {1 2 3 4};
print a; 

b = {1,2,3,4};
print b;

ab = {1 2, 3 4, 5 6};
print ab;

/* In SAS, matrix[row #, col#] */
ab[2,2] = 8 ; 
print ab;

ab[,2] = 3; 
print ab;

/* Function j() -- j(# rows, # cols, value to populate with) */
thebestvectorever = j(50,2,1);
print thebestvectorever;


m = j(4,1,.);

do i=1 to nrow(m);
	m[i,] = i ;
end;
print m;

/* (1) Create a 4x4 matrix
   (2) Run a two-level loop that
		populates this matrix with
		the index number
*/

a = j(4,4,.);

do i=1 to nrow(a);
	do j=1 to ncol(a);
		a[j,i] = i*j;
		print a;
	end;
end;

r = uniform(12345);

rm = j(4,4,.);
do i=1 to nrow(rm);
	do j=1 to ncol(rm);
		rm[i,j] = uniform(12345);
	end;
end;
print rm;

/* Repeat above, but ensure that no value
	is above 0.5 */
rm = j(4,4,.);
do i=1 to nrow(rm);
	do j=1 to ncol(rm);
		rm[i,j] = uniform(12345);
		if rm[i,j]>0.5 then j=j-1;
	end;
end;
print rm;

k = j(2,1,.);
call randgen(k,"Uniform");

g = mean(k);

/* Generating means from uniform distribution */
 /* Initiate holding matrix */
	hold = j(100,1,.);

	do i=1 to nrow(hold);
		t = j(3,1,.);
		call randgen(t,"Uniform");
		t_mean = mean(t);
		hold[i,] = t_mean;
	end;
	print hold;

	mean_mean = mean(hold);
	print mean_mean;


/* Concatenating vectors/matrix */
	one = j(3,1,1);
	gg = {3 1, 2 1, 3 2};
	jj = {5, 6, 7};

	/* Horizontal concatenation */
	h_conc = one || gg;
	print one gg;
	print h_conc;

	/* Vertical concatenation */
	v_conc = one // jj;
	print v_conc;