*==========================================================================*
* randomly split the stock_hrs_2006 dataset into 5 folds for 5-fold CV  
*   
*   Yifan Wei
*   09/01/2020
*==========================================================================*

include "common.do"
log using ID_selection_stock_hrs_2006_5fcv.log, replace

use "../../../input_data/stock_hrs_2006.dta", clear
keep hhidpn 
count 

set seed 2020

gen rand = runiform()
sort rand

gen fold = .
replace fold = 1 if rand<=0.2&rand>0
replace fold = 2 if rand<=0.4&rand>0.2
replace fold = 3 if rand<=0.6&rand>0.4
replace fold = 4 if rand<=0.8&rand>0.6
replace fold = 5 if rand<1&rand>0.8
tab fold

keep hhidpn fold 
save "../../../input_data/5fcrossvalidation_w8.dta", replace


use "../../../input_data/stock_hrs_2006.dta", clear
merge 1:1 hhidpn using "../../../input_data/5fcrossvalidation_w8.dta"
keep if fold==1
saveold "../../../input_data/stock_hrs_2006_5fcv_1.dta", version(11) replace

use "../../../input_data/stock_hrs_2006.dta", clear
merge 1:1 hhidpn using "../../../input_data/5fcrossvalidation_w8.dta"
keep if fold==2
saveold "../../../input_data/stock_hrs_2006_5fcv_2.dta", version(11) replace

use "../../../input_data/stock_hrs_2006.dta", clear
merge 1:1 hhidpn using "../../../input_data/5fcrossvalidation_w8.dta"
keep if fold==3
saveold "../../../input_data/stock_hrs_2006_5fcv_3.dta", version(11) replace

use "../../../input_data/stock_hrs_2006.dta", clear
merge 1:1 hhidpn using "../../../input_data/5fcrossvalidation_w8.dta"
keep if fold==4
saveold "../../../input_data/stock_hrs_2006_5fcv_4.dta", version(11) replace

use "../../../input_data/stock_hrs_2006.dta", clear
merge 1:1 hhidpn using "../../../input_data/5fcrossvalidation_w8.dta"
keep if fold==5
saveold "../../../input_data/stock_hrs_2006_5fcv_5.dta", version(11) replace


log close