CountryData SDMX for Development Indicators

CountryDataSDMX for

Development Indicators

MDG DSD vs. the Di Database:

Using the Mapping Tool

MDG Data Structure Definition (DSD) background

Developed by SDMX Task Team of IAEG on MDGs

Supports exchange of MDG Indicator data between international agencies (UN, UNICEF, UNESCO, …) Implemented in SDMX 2.0 Latest version (2.4) finalised in Feb 2013

DevInfo (Di) background

Data dissemination software supported

and promoted by UNICEF DevInfo7 (Di7) launched in Nov 2012

SDMX 2.1 & 2.0 compliant Web base software 9 out of 11 project countries using

DevInfo Stable version compare to previous

releases

Simple relation between Di & DSDDi Database MDG DSD

• Area

• Indicator

• Unit

• Subgroup (i.e. Sex, Age, Location etc.)

• Source

• Time Period

• Footnotes

• Frequency (Default = “Annual”)

• Reference Area

• Series

• Units of measurement

• Unit multiplier (Default = 0)

• Location (Default = “Total”)

• Age group (Default = “All Ages”)

• Sex (Default = “Both Sexes”)

• Source Type (Default = “NA”)

• Source details

• Time Period

• Time period details

• Nature of data points (Default = “C”)

• Footnotes

Mapping to the DSD DSD dimensional structure means

values are mandatory for LOCATION, SEX & AGE GROUP.

Due the nature of this domain (i.e. MDGs), not obvious which values should be used in these dimensions

For example, what is SEX for “Births attended by skilled personnel”:

Not Applicable? Total? Female?

Mapping to the DSD

Inconsistent mappings lead to duplications and other anomalies

In CountryData, mappings for indicators/ time series are agreed before data exchange (see mapping for MDGs from 1st workshop)

However, this is just one side of the story…

Mapping to the DSD

Understanding the structure and contents of the origin database is fundamental to the mapping process

Mapping to the DSD requires the data to enter into certain ‘restrictions’ it’s not bound by in the database (and vice versa).

Mapping to the DSD The mapping tool in di software is

designed to work with the di database as simply as possible…

the tool is based on mapping between the codelists of the DSD and origin database; certain situations require some further manual effort to map a time series; and sometimes a “fix” is required to the database where the data simply isn’t valid or it’s duplicated.

Therefore it’s good to review di structure to understand where these issues usually occur.

Area, hierarchical dimension IUS = Indicator, Unit and Subgroup

Time series data are stored with the combination of the 3 dimensions Indicator Unit Subgroup: Combination of one or more sub-dimensions

Source & Time Period Together with IUS “uniquely” defines each data value

Footnote “Free text” field stored with data value

Di Data Architecture

IUS: Indicator Unit Subgroup Indicator, for example:

Infant Mortality AIDS Death Malaria Death

Similar to SERIES in the DSD Contains only Indicator specific values

Di INDICATOR

IUS: Indicator Unit Subgroup Unit:

Percentage Number USD Square KM

Similar to UNIT of Measurement in DSD Contains only Unit specific values

Di UNIT

IUS: Indicator Unit Subgroup SubGroup Dimension:

Combination of one or more sub-dimensions

“Age”, “Sex”, “Location” and “Other” sub-dimensions are set initially in database

Specific values can be created under each sub-dimension

Relate to SEX, AGE GROUP and LOCATION in DSD.

Di SUBGROUP

IUS: Indicator Unit Subgroup Formation Logic:

Sub-Dimension Age Sex Location Other

Sub-Dimension values

< 1 Year< 5 Year

5 – 10 Year

MaleFemale

UrbanRural

TotalRice

Wheat

SUBGROUP(Combination)

<1 Year Male<5 Year Female RuralUrban

Di SUBGROUP

Di SUBGROUP

Di SUBGROUP

Di SOURCE

Di TIME PERIOD

Once data exists in di7 web-based software then data can be mapped and published which conforms with the MDG DSD.

This is all done online through the di7 web-based repository through the administration profile, so let’s begin…

Di Mapping Tool: Introduction

Getting Started…

Scroll down to ‘Registry’ menu

Log onto administrative profile

Log onto administrative profile

Full access to ‘Registry’ features

Prepare the Dbase for mapping

Prepares the SDMX artefactes

Ready to ‘Upload’ the DSD

Choose a DSD from your folders

Choose a DSD from your folders

DSD Upload is a success…

Now you are ready to map…

1st Step: Codelist mapping

SEX CodeListNA Not applicableF FemaleM MaleT Both sexes

UNIT CodeList

NA Not applicable

CUR_LCU Local currency

USD USD

NUMBER Number

RATIO Ratio

PERCENT Percent

KM2 Square kilometers

T Metric Tons

PER_100_LIVE_BIRTHS Per 100 live births

PER_100_POP Per 100 population

PER_1000_LIVE_BIRTHS Per 1,000 live births

PER_1000_POP Per 1,000 population

PER_100000_LIVE_BIRTHS Per 100,000 live births

PER_100000_POP Per 100,000 population

AGE CodeList NA Not applicable000_099_Y All age ranges000_006_M under 6 month olds000_005_Y under 5 year olds000_001_Y under 1 year olds000_018_Y under 18 year olds000_006_Y under 6 year olds010_005_Y 10-14 year olds015_005_Y 15-19 year olds015_010_Y 15-24 year olds015_035_Y 15-49 year olds006_054_M 6-59 months old006_009_Y 6-14 year olds005_013_Y 5-17 year olds015_050_Y 15-64 year olds

Location CodeListT T Total (national level)U U UrbanR R Rural

Indicator CodeList

SH_HIV_INCD HIV incidence rate

SH_MLR_MORT Notified cases of malaria

SE_ADT_1524 Literacy rate

SE_PRM_CMPL Primary completion rate

1st Step: DSD Codelists

1st Step: (A) Map Indicator codes

1st Step: (B) Map Unit codes

1st Step: (C) Map Subgroup codes

1st Step: (C) Choose Subgroup list

1st Step: (C) Map Age subgroup

1st Step: (C) Map Sex & Location

1st Step: (D) Map Area

1st Step: Save codelist mappings

1st Step: Ignore warning

1st Step: Confirm mapping saved

Exercise 1: Codelist mapping Use unstats.un.org/unsd/demodiweb[1-6] Username = webmaster@xyz.com Password = support@2012 Map the codelists (where possible) for

Unit Age Sex Location Area And just one indicator, “Antenatal care coverage for

at least one visit”

1st Step: Complete

2nd Step: Confirm IUS mapping

2nd Step: Save IUS Mappings

Exercise 2: mapping time series Use unstats.un.org/unsd/demodiweb[1-6] Username = webmaster@xyz.com Password = support@2012

Map the time series for1. “Antenatal care coverage for at least four visits”

2. “Employment to population ratio”

3. “Literacy rate of 15-24 year-olds”

4. “Death rate associated with malaria”

5. “Proportion of population using solid fuels”

2nd Step: Complete

Final Step: Register the mappings

Final Step: Select mappings

Final Step: Generate SDMX-ML

Final Step: Complete

Exercise 3: Publish time series Use unstats.un.org/unsd/demodiweb[1-6] Username = webmaster@xyz.com Password = support@2012

Publish/ register the time series for1. “Antenatal care coverage for at least four visits”

2. “Employment to population ratio”

3. “Literacy rate of 15-24 year-olds”

4. “Death rate associated with malaria”

5. “Proportion of population using solid fuels”

Why the 2nd step?

The default values for SEX, LOCATION or AGE GROUP mapping may not be applicable to all mappings

The codelist mapping may only provide a partial mapping of the time series (i.e. more information is required)

These changes are made in the 2nd step.

This is all done online through the di7 web-based repository through the administration profile, so let’s begin…

Where are the default values?

Admin panel: Application settings

Insert screens shot/details of admin panel and default value storage…

Application settings has all mapping default values

Antenatal care coverage for at least one visit - PercentSex

Location Rural UrbanAge 15-49 yr 15-49 yr 15-49 yr

Time Period2000 95.6 92.2 96.52004 96.5 95.9 96.32006 98.8 98.7 98.62010 98 97.4 98.1

Female

SERIES LABEL UNITS LOCATION AGE_GROUP SEX

Antenatal care coverage, at least four visits PERCENT T 015_035_Y F

Antenatal care coverage, at least one visit PERCENT T 015_035_Y F

Manual mapping of SUBGROUP

Indicator Unit

• Where a subgroup value is missing the default values will apply, for example…

?

Default Values• Location = T• …

SERIES LABEL UNITS LOCATION AGE_GROUP SEX

Antenatal care coverage, at least four visits PERCENT T 015_035_Y F

Antenatal care coverage, at least one visit PERCENT T 015_035_Y F

Antenatal care coverage rate - PercentLocation Total Rural Urban

Time Period2000 44.8 41.2 67.32005 71.8 70.4 80.72010 89.1 87.6 97

Manual mapping of SUBGROUP

Indicator Unit Subgroup for Age and Sex?

Default Values• …• Age Group = 000_099_Y• Sex = T

• So subgroups coverage affects the number of manual changes which have to be made…

Land under forest cover - PercentTime Period Data Value

1993 59.821997 58.62002 61.152005 602006 59.092008 57.992009 57.62010 57.56

SERIES LABEL UNITS LOCATION AGE_GROUP SEX

Land area covered by forest PERCENT T NA NA

Manual mapping of SUBGROUP

Indicator Unit Subgroups?

Default Values• Location = T• Age Group = 000_099_Y• Sex = Both sexes

• Common example of where default subgroup mapping do not apply

SERIES LABEL UNITS LOCATION AGE_GROUP SEX

Adolescent birth rate PER_1000_POP T 015_005_Y F

Adolescent birth rate - Births per womanLocation Rural Total Urban

Age 15-19 yr 15-19 yr 15-19 yrTime Period

1995 1141999 772003 69.42005 54.6 59.8 63.2

2008-2009 81 67 62

Manual mapping of SUBGROUP

Indicator Unit

Subgroup for Sex?

Default Values• …• …• Sex = T

• Common example of where default subgroup mapping do not apply

• Common example of where default subgroup mapping do not apply

SERIES LABEL UNITS LOCATION AGE_GROUP SEX

Children sleeping under insecticide-treated bed nets PERCENT T 000_005_Y T

Children under-five sleeping under insecticide-treated net (ITN) - Percent, 2011Other Data ValueRural 45.9Total 39Urban 30.1

Manual mapping of SUBGROUP

Indicator Unit

Subgroup for Location, Age and Sex?

Default Values• Location = T• Age Group = 000_099_Y• Sex = T

?

Condom use at last high-risk sex - PercentLocation Total

Sex Female Male Female Male Female MaleAge 15-24 yr 15-24 yr 15-24 yr 15-24 yr 15-24 yr 15-24 yr 15-24 yr

Other Total 15-24 yrTime Period

2000-2001 49.8 39 612002-2003 55.12004-2005 48 53 47.7 51.3 66.9 70.52005-2006 52.9 27.1 65.3

2006 38.3 54.52012 70 73

Rural Urban

SERIES LABEL UNITS LOCATION AGE_GROUP SEX

Condom use at last high-risk sex PERCENT T 015_010_Y M

Condom use at last high-risk sex PERCENT T 015_010_Y F

Manual mapping of SUBGROUP

Indicator Unit

• If the subgroups are sorted more simply, this also helps with the mapping:

?

Default Values• Location = T• …

SERIES LABEL UNITS LOCATION AGE_GROUP SEX

Condom use at last high-risk sex PERCENT T 015_010_Y M

Condom use at last high-risk sex PERCENT T 015_010_Y F

Condom use at last high-risk sex - Percent

OtherFemale 15-24 yr

Male 15-24 yr

Rural female 15-24 yr

Rural male 15-

24 yr

Urban female 15-24 yr

Urban male 15-

24 yrTime Period

2003 32.7 51.7 21.2 38.9 32.5 50.32008 29.2 46.3 21.6 40.6 33.5 52

• Common example of where default subgroup mapping do not apply

Manual mapping of SUBGROUP

Indicator Unit

Subgroup for Location, Age and Sex?

Default Values• Location = T• Age Group = 000_099_Y• Sex = T

?

Back to mapping…

2nd Step: Amend IndicatorWhen using the check box to tick the mapping, you are “fixing” the mapped DSD values. If the box is unchecked again and the mappings saved, then DSD values revert to those mapped at codelist/ default values (i.e. any manual changes are undone.)

Final Step: Register new mappings

Exercise 4: Amend time series Use unstats.un.org/unsd/demodiweb[1-6]

Map/ amend/ publish the time series for;1. “Antenatal coverage rate”

2. “Children orphaned by AIDS”

3. “Children under-five sleeping under insecticide-treated net (ITN)”

4. “Proportion of births attended by skilled health personnel”

5. “Share of women in wage employment in the non-agricultural sector”

6. “Proportion of urban population living in slums”

More complex mappings under the 1st and 2nd mapping step?

The most common changes made to mappings are between subgroups and the Sex, Age Group and Location dimensions

But sometimes manual changes are required between di and DSD indicator and unit, either… More than one di code relates to a single DSD code

OR More than one DSD code relates to a single di code

IndicatorTime

PeriodData

ValueProportionof seats headed by women in

national parliament - Percent1999

11.1Seats held by men in national parliament -

Number2008

80Seats held by women in national

parliament - Number2008

14Seats in national parliament - Number 2008 94

SERIES LABEL UNITS LOCATION AGE_GROUP SEX

Seats in national parliament NUMBER T 000_099_Y MSeats in national parliament NUMBER T 000_099_Y FSeats in national parliament PERCENT T 000_099_Y F

Many-to-one mapping for Indicator codelist (Example 1)

Indicator

Indicator

Indicator

OtherFemale 15-24 yr

Male 15-24 yr

Total

Indicator2003 72.2 75.12008 65.9 76.1

Men 15-24 years with comprehensive knowledge of AIDS - Percent

200834.2

Women 15-24 years with comprehensive knowledge of AIDS - Percent

200828.3

Population 15-24 year-olds who have comprehensive correct knowledge of HIV/AIDS - Percent

Time Period

SERIES LABEL UNITS LOCATIONAGE_GROUPSEX

Population with comprehensive correct knowledge of HIV/AIDS PERCENT T 015_010_Y M

Population with comprehensive correct knowledge of HIV/AIDS PERCENT T 015_010_Y F

Many-to-one mapping for Indicator codelist (Example 2)

Indicator

Indicator

SERIES LABEL UNITS LOCATION AGE_GROUP SEX

Population below national poverty line PERCENT R 000_099_Y T

Population below national poverty line PERCENT T 000_099_Y T

Population below national poverty line PERCENT U 000_099_Y T

Location Total Rural UrbanIndicator Time Period

1993 51.22000 60.3 65.7 14.3

2005-2006 56.7 61.9 28.52010 44.9 48.7 22.12001 60.42005 56.92006 56.92008 56.9

Population below national poverty line - Percent

Share of population below poverty line - Percent

Indicator Indicator

Many-to-one mapping for Indicator codelist (Example 3)

SERIES LABEL UNITS LOCATION AGE_GROUP SEX

Gender Parity Index in primary level enrolment RATIO T 000_099_Y T

Other Total Rural UrbanIndicator Time Period

Gender parity index in primary education - Ratio 2002 0.922003 0.922004 0.932005 0.932006 0.962007 0.96

Gender parity index at primary education - Index 2007-2008 0.962008 0.99 0.99 0.992009 0.96

Unit Unit

Many-to-one mapping for Unit codelist (Example 1)

SERIES LABEL UNITS LOCATION AGE_GROUP SEX

Unmet need for family planning, limiting PERCENT T 015_035_Y F

Unmet need for family planning, spacing PERCENT T 015_035_Y F

Unmet need for family planning PERCENT T 015_035_Y F

Unmet need for family planning - PercentLocation Total Rural Urban

Other Limiting Spacing

1992 19.4 21 40.42000 11.6 24 35.62005 13.4 24.5 37.9 38.4 34.42010 18.9 19.5 15.5

Time Period

Manual mapping of INDICATOR

Indicator Unit

Manual change

?

SERIES LABEL UNITS LOCATION AGE_GROUP SEX

Mobile cellular telephone subscriptions NUMBER T NA NATelephone lines NUMBER T NA NA

Telephone lines - NumberLocation Total

Telephone Cellular lines Fixed lines

2000 72,602 58,2612001 276,034 56,1472002 505,627 59,4722003 893,035 65,7932004 1,165,035 82,4952005 1,525,125 100,7772006 2,697,616 129,863 2,827,4792007 5,163,414 165,788 5,329,2022008 8,554,864 168,481 8,723,3452009 9,383,734 233,533 9,617,2672010 12,828,264 327,114 13,155,378

Time Period

•

Manual mapping of INDICATOR

Indicator

Unit

Manual change

?

SERIES LABEL UNITS LOCATION AGE_GROUP SEX

Gender Parity Index in primary level enrolment RATIO T 000_099_Y T

Gender Parity Index in primary level enrolment - PercentTime

PeriodData Value

2000 0.932003 0.972006 0.992009 1

Manual mapping of UNIT

Indicator Unit

•

Manual change• Unit = “Ratio”

Back to mapping…

1st Step: many di to 1 DSD code

2nd Step: 1 di to many DSD codes

Final Step: Register new mappings

Exercise 5: Complex time series Use unstats.un.org/unsd/demodiweb[1-6]

Map/ amend/ publish the time series for;1. “Contraceptive prevalence rate”

2. “Primary completion rate”

3. “Gender parity index in primary education”

4. “Seats held by men in national parliament”

5. “Seats held by women in national parliament”

6. “Telephone lines”

Other issues encountered with generating SDMX from DevInfo

Type NameDimension SeriesDimension Units of measurementDimension LocationDimension Age groupDimension SexDimension Reference AreaDimension Time Period

• The MDG DSD requires any data point to be uniquely described by the following dimensions;

• However, DevInfo allows data to be stored in overlapping time intervals and with multiple sources. These issues need to be resolved to conform to the “uniqueness” required by the MDG DSD.

Multiple sources

Proportion of population with access to improved sanitation - PercentLocationSource CPC Census 1995 NCEHWS_2007 NCEHWS 2003 NCEHWS 2004

Time Period1990 111995 292000 372001 402002 42 41.62003 42 42.22004 44 44.32005 462006 472007 49

Total

SERIES LABEL UNITS LOCATION AGE_GROUP SEX

Population using improved sanitation facilities PERCENT R 000_099_Y T

Population using improved sanitation facilities PERCENT T 000_099_Y T

Population using improved sanitation facilities PERCENT U 000_099_Y T

Allowable in DevInfo but not in the DSD

Overlapping time

Infant mortality rate - Deaths per 1000 live birthsLocation Total

Time Period1990-1994 27.31995-1999 25.51999-2003 24.22005-2006 25.32005-2009 18.9

SERIES LABEL UNITS LOCATION AGE_GROUP SEX

Infant mortality rate PER_1000_LIVE_BIRTHS T 000_001_Y T

• This issue is only a problem where overlapping periods begin from the same year, as the mapping tool takes the first year in the period as the value for the “Time Period” dimension.

Targets in the database

Targets are also an issue when found in the database since they should not be exchanged as observed values

Target in database (Example 1)

Proportion of people living below the national poverty line - PercentLocation Total Rural Urban

Other MDG target

1990 481992 46 51.8 26.51997 39.1 42.5 22.12002 33.5 37.6 19.72015 24

Time Period

SERIES LABEL UNITS LOCATION AGE_GROUP SEX

Population below national poverty line PERCENT R 000_099_Y T

Population below national poverty line PERCENT T 000_099_Y T

Population below national poverty line PERCENT U 000_099_Y T

Sometimes stored as subgroup which can be ignored at the 2nd stage…

Location Rural Urban

SourceCPC

Census CPI Census

2005Governme

nt 2007SPC RHS

2000SPC RHS

2000SPC RHS

2000

1995 6502000 530 580 1702005 4052015 260

Time period

TotalMaternal mortality ratio - Deaths per 100,000 live births

SERIES LABEL UNITS LOCATION AGE_GROUP SEX

Maternal mortality ratio PER_100000_LIVE_BIRTHS T 000_099_Y F

Target in database (Example 2)

But other times can be found as a time period among observed values…

Use of filters at registration

To deal with the issues of; multiple sources for a given time period, overlapping time period beginning at the same

year; And targets presented alongside observed values

The mapping tool provides a feature to filter out data from a generated SDMX message associated with specific time periods and source references.

Back to mapping…

Final Step: Filter by time/ source

Final Step: Select source filter

Final Step: Select time filter

Final Step: Register new mappings

Final Step: Complete

Exercise 6: Filter time series Use unstats.un.org/unsd/demodiweb[2-6]

Map/ amend/ publish the time series for;1. “Under-five mortality rate”

2. “Maternal mortality ratio (MMR)”

3. “Net enrolment ratio in primary education (NER)”

4. “Orphans primary school enrolment”

5. “Tuberculosis prevalence rate”

6. “Proportion of the population using improved sanitation facilities”

DSD Maintenance

• The mapping and registry tool allows users to edit and delete the DSD as well as upload.

• For when the DSD is updated, it is recommended to edit the DSD rather than delete

• DSD deletion has the effect of removing all the mappings and subscriptions used for that DSD

DSD Maintenance…

DSD Header…