clear all
cd c:\chililabdata
use individres
sort individid res_sdate
gen thisround = substr(res_sobserveid, 1,2)
drop res_episodeid  res_seventtype res_sobserveid res_eobserveid res_sobserveidGuid res_eobserveidGuid
by individid: gen p=_n
by individid: gen last=_N
cd data
save individres_base, replace
keep if p==1
duplicates drop individid, force
gen sdate = res_sdate/86400000
keep individid  sdate thisround
format sdate %td
sort individid
save beginRes, replace

use  individres_base
keep if p == last
gen edate = res_edate/86400000
replace edate = mdy(12,31,2013) if edate ==.
rename res_eeventtype event
format edate %td
keep individid  locationid edate event
duplicates drop individid, force
sort individid
merge 1:1 individ using beginRes
drop _merge
gen peryear = (edate-sdate)/365.25

gen endpoint =0
replace endpoint =1 if event == "DTH"

gen year = year(edate)
gen h = halfyear(edate)
gen round = yh(year,h)-89
tostring round, generate(rounds)
replace rounds = "0" + rounds if strlen(rounds) ==1
gen individkey = individid + rounds
drop rounds h round
sort individkey
cd death
merge 1:1 individkey using rMembership, keepusing(socialgpid)
sort individid
drop _merge

/*get gender*/
merge m:1 individid using individual, keepusing(gender birthdate)

gen birthyear = year(birthdate/86400000)
keep if _merge ==3
drop _merge


/*get Education*/
merge 1:m individkey using rEducation, keepusing(eduid)
keep if _merge ==3
drop _merge

/*get Occupation*/
merge 1:m individkey using rOccupation, keepusing(occupationid)
keep if _merge ==3
drop _merge

/*get Marital Status*/
merge 1:m individkey using rMarstatus, keepusing(marstatusid)
keep if _merge ==3
drop _merge
tostring year, replace
destring gender , replace
rename gender sex
gen socialkey = socialgpid + year
sort socialkey
merge m:1 socialkey using houseinfo, keepusing(quintile)
keep if _merge ==3
drop _merge
gen bod = birthdate/86400000
drop event year individkey socialgpid birthdate socialkey
sort individid

merge 1:1 individid using 60inds, keepusing(individid) 
keep if _merge==3
drop _merge
gen eAge  = ceil((edate-bod)/365.25)
gen sAge = ceil((sdate -bod)/365.25)
format bod %td
gen date60 = mdy(month(bod), day(bod), year(bod) + 60)
format date60 %td
replace peryear = (edate -date60)/365.25 if date60>sdate
gen duration = round(peryear)
drop if duration <= 0
drop if duration > 10
destring occupationid , replace
destring edu, replace
drop if edu==0
gen communeid = substr(locationid, 1,2)
destring communeid, replace
recode eduid (1/3 = 1) (8/11 = 1) (12/14 = 2) (15/19 = 3) (4/7=4) 
recode occupationid (1= 1) (6 = 2) (5 = 3) (2/4=4) (7/13=4)
recode marstatusid  (0=0)(1 6 = 1) (2/4 = 2) (5=3) 
recode communeid (1 = 1) ( 2= 2) (3=1) (4= 2) (5 =2) (6 = 1) (7 =2), generate(region)
label define Region 1 "Đô thị" 2 "Nông thôn"
label values region Region

label  variable marstatusid "Tinh trang hon nhan"
label  variable eduid "Trinh do hoc van"
label  variable occupationid "Nghe nghiep"
label  variable sex "Gioi tinh"
label  variable quintile "Muc do giau ngheo"
label variable communeid "Xã"
label define sexes 1 "Nam" 2 "Nu"
label values sex sexes
label define Occupation 1 "1. Làm ruong" 2 "2. Huu tri" 3 "3. Buon ban dich vu" 4 "4. Khac" 
label values occupationid Occupation

label define edu 1 "1. Duoi THCS"  2 "2. THSC" 3 "3. THPT" 4 "4. Tren THPT"   
label values eduid edu

label define marstatus 0 "0. Khong biet" 1 "1. Song cung vo/chong" 2 "2. Ly than/Ly di/Goa"  3 "3. Doc than"
label values marstatusid marstatus
label define CommnueLabel 1 "Sao Đỏ" 2 "An Lạc" 3 "Phả Lại" 4 "Văn An" 5 "Lê Lợi" 6 "Bến Tắm" 7 "Hoàng Tiến"
label values communeid CommnueLabel

saveold indperiod, replace