DOE_HW1_update_1

/*Design of Experiments: Homework 1*/

/*IMPORT DATA*/
libname zipline '\\vmware-host\Shared Folders\Desktop\SAS Programming\SAS Datasets';

data zip;
	set zipline.rduch (drop=VAR1);
	/*Creating distance variables to each location*/
	distloc1=sqrt(((long--78.878130)**2)+((lat-35.89314)**2));
	distloc2=sqrt(((long--78.875880)**2)+((lat-35.74628)**2));
	distloc3=sqrt(((long--78.676540)**2)+((lat-35.7724)**2));
	distloc4=sqrt(((long--79.054280)**2)+((lat-35.90535)**2));
	distloc5=sqrt(((long--78.575981)**2)+((lat-35.86696)**2));
	/*Find minimum distance*/
	closest_dist=min(distloc1, distloc2, distloc3, distloc4, distloc5);
	/*Create variable closest_loc to assign the closest location*/
	if closest_dist=distloc1 then closest_loc='Location 1';
	else if closest_dist=distloc2 then closest_loc='Location 2';
	else if closest_dist=distloc3 then closest_loc='Location 3';
	else if closest_dist=distloc4 then closest_loc='Location 4';
	else closest_loc='Location 5';
	/*Create distance class - these do not need to be equally spaced*/
	if closest_dist lt 0.05 then distance_class='Closest';
	else if 0.05 ge closest_dist lt 0.1 then distance_class='Close';
	else if 0.1 ge closest_dist lt 0.15 then distance_class='Medium';
	else if 0.15 ge closest_dist lt 0.2 then distance_class='Far';
	else distance_class='Farthest';
	/*Create weight variable for sampling*/
	weight=1/closest_dist;
run;

/*DATA EXPLORATION*/
proc univariate data=zip;
	var closest_dist;
	histogram closest_dist / normal(color=red);
run;

proc univariate data=zip;
	var lat long;
	histogram lat / normal(color=red);
	histogram long / normal(color=blue);
run;

proc means data=zip;
	class closest_loc;
	var closest_dist;
run;
quit;

proc print data=zip (obs=10);
run;

proc sgplot data=zip;
  vbar closest_loc / stat=percent dataskin=crisp;
  xaxis display=(nolabel noticks);
  yaxis grid;
  title 'Closest Locations by Percent';
run;
title;
/*Should we aim to get representative people by location?*/

/*PROC POWER;
	ONESAMPLEMEANS TEST=t
	NULLMEAN=200 
	MEAN=225     
	STDDEV=20     
	ALPHA=0.05
	POWER=0.9
	NTOTAL=.;
RUN;*/
/*-------------------------------------------------------------------*/
/*Simple Random Sample*/
proc surveyselect data=zip
	method=srs n=8000 out=SRS;
	title 'Simple Random Sample';
run;

/*-------------------------------------------------------------------*/
/*Stratified Random Sample - Location*/
/*Sort into Location*/
proc sort data=zip;
	by closest_loc;
run;

/*Sample*/
proc surveyselect data=zip
	method=srs n=1600 
	seed=8675309 out=StrataSample;
	strata closest_loc;
run;


/*-------------------------------------------------------------------*/
/*Stratified by Location Proportionally*/
proc surveyselect data=zip
	out=strat_samp_prop
	seed=8675309
	method=srs
	sampsize=8000;
	strata closest_loc / alloc=prop;
run;