Date post: | 13-Dec-2015 |
Category: |
Documents |
Upload: | amie-pitts |
View: | 222 times |
Download: | 0 times |
2
Overview
• Introduction to the BmTS
• 3 key themes:1. People, process, methods…then
software
2. Advances in Stats NZ methodology
3. Evolutionary change - business, cultural & programme
• Questions
4
BmTS Deliverables
1. Standard processes• 80/20
2. Disciplined approach• Data and metadata
3. Enterprise-wide technical architecture
8
Business Process Model
Need Build Collect Process AnalyseDesign Disseminate
Establishpopulation
Generatesample
ValidateAnd Q.A.
Maintainsample
Identify sample
Manageproviders
Setupcollection
RunCollection
Loaddata
9
Business Process Model
Establishpopulation
Generatesample
ValidateAnd Q.A.
Maintainsample
Identify sample
Manageproviders
Setupcollection
RunCollection
Loaddata
Need Build Collect Process AnalyseDesign Disseminate
BDSS Business Process
Ed
itor
An
aly
stS
up
erv
iso
rS
yste
m
Create Base Version
Extract Data &Apply
Derivations,imputations
Produce report
Manual EditingMake changes to existing data and add new records
No
Set Status of Dataset
Yes
Create Analytical Version
Analyse Dataset
Generate Analytical Report
Create Final Version
Yes
Generate Final Dataset
No
Delete Dataset
Generate Report
Create Output
Re-runDerivations,imputations
Set Base Version as Clean Version
Generate Report
Reset StatusTo Base
No
Yes
Set Status to Base
Generate Report
Assess Quality of Dataset
Discard Dataset
YesSet StatusTo Discard
Generate Report
Generate Output Files
Set Final to Published
Generate Report
Set Final to Published
Generate Report
Establishpopulation
Generatesample
ValidateAnd Q.A.
Maintainsample
Identify sample
Manageproviders
Setupcollection
RunCollection
Loaddata
Need Build Collect Process AnalyseDesign Disseminate
10
Business Process Model
Establishpopulation
Generatesample
ValidateAnd Q.A.
Maintainsample
Identify sample
Manageproviders
Setupcollection
RunCollection
Loaddata
Need Build Collect Process AnalyseDesign Disseminate
11
Business Process Model
Build Collect ProcessDesign
CORPORATE
STATISTICAL
MANAGE
Current generic BPM (gBPM)
Methodology
Need Analyse Disseminate
12
Business Process Model
Need Design Collect Process AnalyseBuild Disseminate
CORPORATE
STATISTICAL
MANAGE
Future gBPM
Methodology
13
Process - Progress & successes
• gBPM - for all collections - developed, agreed and used
• Detailed business processes - documented for – Collect, Analyse, Disseminate– Administrative data, data integration, & feasibility projects
14
Methods – Establishment surveys
Advances in:
• longitudinal Business Frame
• size measures on our Business Frame
• modelled tax data for the "small" strata
• regular reselection
• record linkage methodology
• research into sample rotation
• p% rule for confidentialisation of tables
15
Methods – Case study
generic E&I
Processes
E&I Training
New E&Imethods
Standard E&I tools
E&I Plans
E&I Standards
E&I Strategy
& Principles
Editing & Imputation
16
Methods - Progress & successes
• Standard methods – being developed and/or documented
• Standard tools – examples:– BANFF (Statistics Canada) for editing and
imputation– INTERP (in-house) for benchmarking and
interpolation– QualityStage (IBM) for data integration– GREGWT (ABS) for integrated weighting– X12-ARIMA (US Census Bureau) / SADJ (in-
house)
18
Software evolution
BmTS builds on SProceT foundations
• metadata driven systems
• common look and feel
• re-use of 'best practice'
• availability of management information
• dynamic nature of views
• interactive processing
• fully integrated desktop processing
19
Software evolution
BmTS: The next generation
• not a template that is iteratively improved; not in Lotus Notes
• wider scope - end-to-end; used by all Statistics NZ collections
• generic & standard business processes, methods, tools
• workflows, centralised data & metadata; service oriented architecture (SOA)
20
Collect
Future Software - BmTS ComponentsProcess Analyse Disseminate10. Dashboard / WorkflowNeed
Build
Design
2. Output Data Store
CleanData Data
1. Input Data Store
RawData
RADL
Web
Ou
tpu
t C
ha
nn
els
Mu
lti-Mo
da
l Co
llec
tion
CU
RFS
INFO
S
E-Form
CAI
Imaging
Admin.
Data
Off
icia
l Sta
tistic
s S
yste
m &
D
ata
Arc
hiv
e
SummaryData
‘UR’Data
2. Output Data Envt.1. Input Data Environment
9. Reference Data Stores
7. Respondent Management 8. Customer Management
RA
DL
Web
Ou
tpu
t C
ha
nn
els
Mu
lti-Mo
da
l Co
llec
tion
CU
RFS
INFO
S
E-F
ormC
AI
Imaging
Adm
in.D
ataO
ffic
ial S
tatis
tics
Sys
tem
&
CleanData
AggregateData
RawData
SummaryData
‘UR’Data
Da
ta A
rch
ive
3. Metadata StoreStatistical
Process
Knowledge Base
3. Metadata EnvironmentStatistical
Process
Knowledge Base
4. Analytical Environment
5. Information Portal
6. Transformations
23
BmTS Components - Progress10. Dashboard / Workflow
2. Output Data Store
CleanData
AggregateData
1. Input Data Store
RawData
RADL
Web
Ou
tpu
t C
ha
nn
els
Mu
lti-Mo
da
l Co
llec
tion
CU
RFS
INFO
S
E-Form
CAI
Imaging
Admin.
Data
Off
icia
l Sta
tistic
s S
yste
m &
D
ata
Arc
hiv
e
SummaryData
‘UR’Data
2. Output Data Envt.1. Input Data Environment
9. Reference Data Stores
7. Respondent Management 8. Customer Management
RA
DL
Web
Ou
tpu
t C
ha
nn
els
Mu
lti-Mo
da
l Co
llec
tion
CU
RFS
INFO
S
E-F
ormC
AI
Imaging
Adm
in.D
ataO
ffic
ial S
tatis
tics
Sys
tem
&
CleanData
AggregateData
RawData
SummaryData
‘UR’Data
Da
ta A
rch
ive
3. Metadata StoreStatistical
Process
Knowledge Base
3. Metadata EnvironmentStatistical
Process
Knowledge Base
4. Analytical Environment
5. Information Portal
6. Transformations
AdminData
T’form CustomerCRM
SystemMetadata
Link toAnalytics
Link toPortal
URT’form
CategoryEI
AggregateArea
CleanArea
BF
SASBI
GraphicalAnalysis
TSAnalysis
ConfidT’formOutput
cubes
RMportal
CRM
T’formlibrary Logi+
D/board workflow
IDCImaging
Qstage
Datalab
Statisphere
Tablebuilder
BANFF
CallMgmt
MSExcel
24
Software – Progress & successes
• Strategy & Broad Logical Design for 7/10 BmTS components
• Proof of concept / prototype solutions for:– National Accounts / time series data– dissemination products– unit record data: collect to clean
• Standardised collection phase in production• Fact table approach utilised for all data• Reuse of components is happening • User Interface guide developed and utilised• Service oriented architecture in place
25
Changes and challenges
Cultural change required
• business processes as the driver (not ICT)
• focus on commonalities between business areas
• support for and use of standards
• culture of analysis
26
Changes and challenges
• Ownership - of processes, methods, and tools / software
• Monitoring progress
• Clarifying future statistical architecture
• Impact on data quality
• Determining the impact on specific outputs
27
Lessons learned
• People > process > method >… systems– Collection areas focus on their differences
– Compromise: development vs BAU
– Long-term gain has short-term cost
• Evolutionary transformation: – many minor successes and failures
Do not expect to get it 100% right the first time