/*Tab11_prediction_last_step.sas

*/

libname public 'd:\a_data\proj\size_fun\exports\revision\wrap\public';run;

run;


filename f1_1 'd:\a_data\proj\size_fun\exports\revision\wrap\public\z_output_all_1.asc';
filename f2_1 'd:\a_data\proj\size_fun\exports\revision\wrap\public\z_output_sim_1.asc';
filename f3_1 'd:\a_data\proj\size_fun\exports\revision\wrap\public\z_output_simChina_1.asc';

filename f1_2 'd:\a_data\proj\size_fun\exports\revision\wrap\public\z_output_all_2.asc';
filename f2_2 'd:\a_data\proj\size_fun\exports\revision\wrap\public\z_output_sim_2.asc';
filename f3_2 'd:\a_data\proj\size_fun\exports\revision\wrap\public\z_output_simChina_2.asc';
filename f4_2 'd:\a_data\proj\size_fun\exports\revision\wrap\public\output_sim_China_newpop_2.asc';
filename f_data 'd:\a_data\proj\size_fun\exports\revision\wrap\public\output_dataLQ_2.asc';

* f1_1: naicsindex ea_index LQest_base1 LQest_China LQest_base1_aprox LQest_China1_aprox s x LQsal97;
*  f1_2:  naicsindex ea_index LQest_base2 LQest_China1 LQest_base1_aprox LQest_China2_aprox s x LQsal97 LQestChina_newpop1 LQestChina2_newpop_aprox chinanewcat naics;

*  f5_2: LQ97dat LQ07dat;

*First load in data from primary segment;
data part1_1; infile f1_1;
	input naicsindex ea_index LQest_base1 LQest_China1 LQest_base1_approx LQest_China1_approx s popshare97 LQsal97 ; 
	drop s popshare97 LQsal97 ; *will get from other file;

data part2_1; infile f2_1;
	input LQ1_RTM;
data part3_1; infile f3_1;
	input LQ1_RTM_China;

data dat1; merge part1_1 part2_1 part3_1;

*Next load in data from full model with speciality segments;

data part_data;	infile f_data;
	input LQest97 LQest07;

DATA part1_2; INFILE f1_2;
	input naicsindex ea_index LQest_base2 LQest_China2 LQest_base2_approx LQest_China2_approx s popshare97 LQsal97 LQestChina2_newpop LQestChina2_newpop_approx chinanewcat0 naics0;

data part2_2; infile f2_2;
	input LQ2_RTM;
data part3_2; infile f3_2;
	input LQ2_RTM_China;
data part4_2; infile f4_2;
	input LQ2_RTM_China_Newpop;


data dat2;
merge part_data part1_2 part2_2 part3_2 part4_2;

data naics;
set public.ea_cm97;
by naics; if first.naics;
keep naics naicsindex;


proc sort data=dat2;
by naicsindex;

data dat2;
merge dat2(in=in1) naics; by naicsindex; drop naics0;
if in1;

data t97;
set public.ea_cm97;
est97=est_LM; 
estUS97=estUS_LM;
emp97=emphat_LM;
keep naicsindex ea_index est97 emp97 estUS97;

data t07;
set public.ea_cbp07;
est07=est_cbp; emp07=emphat_cbp;
s07=snorm_cbp/snormUS_cbp;
estUS07=estUS_cbp;
empUS07=emphatUS_cbp;
keep naicsindex ea_index est07 emp07 s07 estUS07;
run;



proc sort data=dat2;
by naicsindex ea_index;

proc sort data=dat1; by naicsindex ea_index;

data dat;
merge dat1 dat2 t97 t07; by naicsindex ea_index;
if s>=.05 and LQsal97>=2;
if est97>0 then LQsize97=(s/est97)/(1/estUS97); 
if est07>0 then LQsize07=(s07/est07)/(1/estUS07); 

if LQest97>0 then difLQest_data=100*(LQest07-LQest97)/LQest97 ;
if LQsize07^=. and LQsize97^=. then difLQsize_data=100*(LQsize07-LQsize97)/LQsize97 ;

if LQest_base1>0 then difLQ1_china_exact=100*(LQest_china1-LQest_base1)/LQest_base1 ;
if LQest_base1_approx>0 then difLQ1_china=100*(LQest_China1_approx-LQest_base1_approx)/LQest_base1_approx ;
if LQest_base1_approx>0 then difLQ1_RTM=100*(LQ1_RTM-LQest_base1_approx)/LQest_base1_approx ;
if LQest_base1_approx>0 then difLQ1_china_RTM=100*(LQ1_RTM_China-LQest_base1_approx)/LQest_base1_approx ;

if LQest_base2>0 then difLQ2_china_exact=100*(LQest_china2-LQest_base2)/LQest_base2 ;
if LQest_base2_approx>0 then difLQ2_china=100*(LQest_China2_approx-LQest_base2_approx)/LQest_base2_approx ;
if LQest_base2_approx>0 then difLQ2_RTM=100*(LQ2_RTM-LQest_base2_approx)/LQest_base2_approx ;
if LQest_base2_approx>0 then difLQ2_china_RTM=100*(LQ2_RTM_China-LQest_base2_approx)/LQest_base2_approx ;
if LQest_base2_approx>0 then difLQ2_china_RTM_newpop=100*(LQ2_RTM_China_Newpop-LQest_base2_approx)/LQest_base2_approx ;

if chinanewcat0 = 1 then chinacat2=1;
if chinanewcat0 = 2 then chinacat2=2;
if chinanewcat0 in (3,4,5,6) then chinacat2=3;
run;


ods html file="d:\a_data\proj\size_fun\exports\revision\wrap\public\Tab11_prediction_last_step.html" style=minimal;


proc tabulate data=dat missing;
class chinanewcat0;
var LQsize97 LQest97 LQest07 difLQest_data ;
table all chinanewcat0, N*f=7.0 mean*f=8.1*(LQsize97 LQest97 LQest07) mean*f=8.0*difLQest_data ;
title 'conditonal on s>=.02 and LQsal97>=2';


proc tabulate data=dat;
class chinanewcat0;
var difLQ1_china difLQ1_RTM difLQ1_china_RTM difLQ2_china difLQ2_RTM difLQ2_china_RTM difLQ2_china_RTM_newpop;
table chinanewcat0, N*f=7.0 mean*(difLQ1_china difLQ1_RTM difLQ1_china_RTM difLQ2_china difLQ2_RTM difLQ2_china_RTM difLQ2_china_RTM_newpop)*f=8.0;
title 'conditonal on s>=.02 and LQsal97>=2';



proc ttest data=dat;
where chinacat2 in (1,3);
class chinacat2;
var difLQest_data ;
title 'compare 1 with 3';

proc ttest data=dat;
where chinacat2 in (2,3);
class chinacat2;
var difLQest_data ;
title 'compare 2 with 3';


run;


ods html close;
run;