process-NEED-EULF-2014.do 7.3 KB
Newer Older
Ben Anderson's avatar
Ben Anderson committed
1
* Script to turn original wide 2014 EULF version of DECC's NEED data into:
2
3
4
* 1. a stata wide form xwave file containing the fixed value variables
* 2. a stata wide form file containing just the yearly consumption variables (linked to 1. via HH_ID)
* 3. a stata long form file containing just the yearly consumption variables (linked to 1. via HH_ID)
Ben Anderson's avatar
Ben Anderson committed
5
6
7
* 4. Create codebooks from the above

* Original data available from: UK DATA ARCHIVE: Study Number 7518 - National Energy Efficiency Data-Framework, 2014
8
* http://discover.ukdataservice.ac.uk/catalogue/?sn=7518
Ben Anderson's avatar
Ben Anderson committed
9
10
11
12
13
14
15
16
17
18
19
20

* Notes:
* This dataset is a sample of just over 4 million households which have had an Energy Performance Certificate from the full NEED 'all dwellings' dataset
* Is this all those who have had an EPC or a random sample of all those who've had an EPC?
* Sample bias is unkown - which kinds of dwellings have an EPC?
* Gcons<year>valid variable has undefined labels: G, L, M = ? Presumably 0 = off gas & V = valid?
* ideally DECC should set missing to -99 to aid re-coding and avoid unpleasant surprises in naive analysis!

* Author: Ben Anderson, Energy & Climate Change, Faculty of Engineering & Environment, University of Southampton
* b.anderson@soton.ac.uk
* (c) University of Southampton

21
* Unless there is a different license file in the folder in which this script is found, the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) license applies
Ben Anderson's avatar
Ben Anderson committed
22
23
24
25
26
27
28
29
30
31
32
33
34
35
* http://creativecommons.org/licenses/by-nc/4.0/

clear all

capture noisily log close _all

* written for Mac OSX - remember to change filesystem delimiter for other platforms
local home "/Users/ben/Documents"
local proot "`home'/Work/Data/Social Science Datatsets/DECC"
* for clam
* local proot "`home'/Work/NEED"
local dpath "`proot'/NEED/End User Licence File 2014/"
* NB this is the 2014 EULF we're using
local ifile "need_eul_may2014"
36
* original data file
Ben Anderson's avatar
Ben Anderson committed
37
local dfile_orig "`dpath'UKDA-7518-stata11/stata11/`ifile'.dta"
Ben Anderson's avatar
Ben Anderson committed
38

39

40
* if you really wanted to you could set up a loop to iterate over a list of sample values to create a set of random sub-samples
41
42
43
* we'll just create a random sub-sample of sample% to make testing models etc easier
* 10 = 10% sample, 50 = 50% sample, 100 = 100% sample etc
local sample 20
Ben Anderson's avatar
Ben Anderson committed
44
45
46
47
48
49
50
51
local samplet "`sample'pc"
local sampleby "EE_BAND PROP_TYPE"

local version "v1.1"
* includes production of % samples which maintain the original dimensions used to 
* produce the EULF samples: EE_BAND PROP_TYPE

*local version "v1"
Ben Anderson's avatar
Ben Anderson committed
52
53
54
55
56

set more off

log using "`dpath'/processed/process-NEED-EULF-2014-`version'-$S_DATE.smcl", replace name(main)

57
58
* use these locals to control what happens (set to 0 to skip the code)
* create codebook & some descriptives
Ben Anderson's avatar
Ben Anderson committed
59
local create_codebook = 0
Ben Anderson's avatar
Ben Anderson committed
60
61
62
* create wide form fixed file with (supposedly) unchanging data & a seperate 'wide' consumption data file for cross-sectional analysis
local create_xwavefile = 1
* create long form file with wave (yearly) data - be careful, this take a long time due to large memory use!
Ben Anderson's avatar
Ben Anderson committed
63
64
65
66
local create_longfile = 1

* load the original file
use "`dfile_orig'", clear
Ben Anderson's avatar
Ben Anderson committed
67
68

if `create_codebook' {
Ben Anderson's avatar
Ben Anderson committed
69
70
	* create original EULF codebook
	* not much point running thid for each % sample although the counts etc reported in the codebook won't match
Ben Anderson's avatar
Ben Anderson committed
71
	log off main
Ben Anderson's avatar
Ben Anderson committed
72
	log using "`dpath'/processed/codebook-NEED-EULF-2014-`version'-`sample'pc-$S_DATE.smcl", replace name(cb)
Ben Anderson's avatar
Ben Anderson committed
73
74
75
76
77
78
79
	desc
	di "** no idea what G, L, M mean in the 'valid' variables - presumably 0 = off gas & V = valid?"
	codebook
	log close cb
	log on main
}

Ben Anderson's avatar
Ben Anderson committed
80
81
82
83
84
85
86
87
***** random sample ****
* select a random sample but ensure proportions of sampleby are kept
di "* Keeping `sample'% sample by `sampleby'"
sample `sample', by(`sampleby')

tab `sampleby', mi


Ben Anderson's avatar
Ben Anderson committed
88
89
if `create_xwavefile' {
	* create the file with data that (notionally) doesn't change
Ben Anderson's avatar
Ben Anderson committed
90

Ben Anderson's avatar
Ben Anderson committed
91
92
93
	* create a wide consumption file
	preserve
		keep HH_ID Gcons* Econs*
94
		compress
Ben Anderson's avatar
Ben Anderson committed
95
		save "`dpath'/processed/`ifile'_consumptionfile_wide_`samplet'.dta", replace
Ben Anderson's avatar
Ben Anderson committed
96
97
98
	restore
	
	
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
	preserve
		drop Gcons* Econs*
		* fix some mis-codings (or lack of coding of missing)
		local vars "E7Flag2012 CWI LI"
		foreach v of local vars {
			destring `v', force replace
			replace `v' = 0 if `v' !=1
			label def `v' 0 "No or N/A" 1 "Yes"
			label val `v' `v'
		}
		
		* turn '99' into missing - ideally missing should be -99 to aid re-coding and avoid unpleasant surprises in naive analysis!
		replace FP_ENG = . if LOFT_DEPTH  == 99
		replace LOFT_DEPTH = . if LOFT_DEPTH  == 99
		
		* what do G, L, M mean in the gas 'valid' variables - presumably 0 = off gas & V = valid?
		
		tabstat IMD_WALES, by(REGION) s(mean min max n)
		* there seem to be some welsh LSOAs allocated to English GORs?
		
		tabstat IMD_ENG, by(REGION) s(mean min max n)
		* there seem to be some English LSOAs allocated to Wales?
Ben Anderson's avatar
Ben Anderson committed
121
	
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
		tabstat FP_ENG, by(REGION)
		* REGION is ONS admin codes
		* create a new variable with meaningful labels
		gen ba_region = 1 if REGION == "E12000001"
		replace ba_region = 2 if REGION == "E12000002"
		replace ba_region = 3 if REGION == "E12000003"
		replace ba_region = 4 if REGION == "E12000004"
		replace ba_region = 5 if REGION == "E12000005"
		replace ba_region = 6 if REGION == "E12000006"
		replace ba_region = 7 if REGION == "E12000007"
		replace ba_region = 8 if REGION == "E12000008"
		replace ba_region = 9 if REGION == "E12000009"
		replace ba_region = 10 if REGION == "W99999999"
		
		lab var ba_region "former Govt Office region (labelled)"
		* http://www.ons.gov.uk/ons/guide-method/geography/beginner-s-guide/administrative/england/government-office-regions/index.html
		lab def ba_region 1 "North East" 2 "North West" 3 "Yorkshire & The Humber" 4 "East Midlands" ///
			5 "West Midlands" 6 "East of England" 7 "London" 8 "South East" 9 "South West" 10 "Wales"
		lab val ba_region ba_region
		
		compress
		log off main
		log using "`dpath'/processed/codebook-NEED-EULF-2014-`ifile'_xwavefile-`version'-`samplet'-$S_DATE.smcl", replace name(cb_xwave)
		desc
		di "** no idea what G, L, M mean in the 'valid' variables - presumably 0 = off gas & V = valid?"
		di "** urban/rural would be helpful"
		codebook
		compress
		save "`dpath'/processed/`ifile'_xwavefile_`samplet'.dta", replace
		log close cb_xwave
		log on main
	restore
Ben Anderson's avatar
Ben Anderson committed
154
155
156
157
158
}

if `create_longfile' {
	* create the long file with as few vars as possible (quicker)
	* still takes a while...
Ben Anderson's avatar
Ben Anderson committed
159

Ben Anderson's avatar
Ben Anderson committed
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
	keep HH_ID Gcons* Econs*

	* panel vars:
	* Gcons2005 Gcons2005Valid Econs2005 Econs2005Valid -> 2012
	local vars "Gcons Econs"
	foreach v of local vars {
		di "* Renaming -> `v'"
		foreach y of numlist 2005/2012 {
			di "* -> `v' (`y')"
			* put year on the end so reshape works
			gen `v'Valid`y' = `v'`y'Valid
			* remove old variable to save time in reshape & space
			drop `v'`y'Valid
		}
	}
Ben Anderson's avatar
Ben Anderson committed
175
	* this takes a LONG time for the full dataset
Ben Anderson's avatar
Ben Anderson committed
176
177
	reshape long Gcons GconsValid Econs EconsValid, i(HH_ID)
	rename _j year
178
	* set as panel
Ben Anderson's avatar
Ben Anderson committed
179
	xtset HH_ID year
Ben Anderson's avatar
Ben Anderson committed
180
	compress
Ben Anderson's avatar
Ben Anderson committed
181
	save "`dpath'/processed/`ifile'_consumptionfile_long_`samplet'.dta", replace
182
	* this leaves us with the long form file in memory
Ben Anderson's avatar
Ben Anderson committed
183
184
185
}

/*
186
* Link xwave data to long form file
Ben Anderson's avatar
Ben Anderson committed
187
* THIS TAKES AGES and creates a 1.5 GB file for the full dataset - use with care
Ben Anderson's avatar
Ben Anderson committed
188
189
190
191
192
193
* now just merge them
* start with long file which may or may not have just been re-created
use "`dpath'/`dfile'_consumptionfile_long.dta", clear

merge m:1 HH_ID using "`dpath'/`dfile'_xwavefile.dta"

Ben Anderson's avatar
Ben Anderson committed
194
save "`dpath'/`dfile'_consumptionfile_long_complete_`samplet'.dta", replace
Ben Anderson's avatar
Ben Anderson committed
195
196
*/

197
198
* done!

Ben Anderson's avatar
Ben Anderson committed
199
log close _all