Standard Process Steps in Statistics

1
Standard Process Steps in Statistics
Robbert Renssen and Astrea Camstra, Statistics
Netherlands
Robbert Renssen rrnn_at_cbs.nl Statistics Netherlands
2
Outline of the Presentation

• Standard Process Steps
• Why
• Future situation
• State of research
• Example a function and a process
• Relation IAF

3
Why

• Management of long-term changes
• From business targets to ideal situation
  (represented by architectural principles)
• From ideal situation to project targets
• From project targets to (physical) solutions
  (e.g. standard tools, process descriptions)

4
High level architecture of SN ?? GSBPM
Design
Chain Management
Statistics Production
Central Storage of data to be exchanged
5
Process modelling in practice

• Process models are very diverse (specific
  problems and solutions, terminology and so on)
• Applications of statistical methods are not
  recognized
• Applications of statistical methods and/or tools
  need preparatory activities
• Re-use of data and mixed mode strategies
  complicate the activities in a process
• Happy and unhappy process flows are entangled

6
Future situation (happy flows)
Data Service Centre
Legend
Alignment
Statistical dataset
Desision step
Standard process step
IT-tool
Business service
7
The state of research

• The current process models of a (large) number of
  processes have been collected (and studied)
• A general framework is formulated
• The process of the Short Term Statistics has been
  modelled in accordance with this framework
• The data validation function is worked out and
  illustrated by practical examples
• The error localisation and matching function are
  under construction (also a few non-statistical
  functions)

8
Example of the data validation function
Edit IF marital status ever married THEN age gt
17

• Age 0,1,2,..120)
• Marital Status (single, ever married)

Q-indicator (passed, failed)
Black Box (statistical method)
Distinction between the nature of a function and
its use domain edits, hard edits, soft edits,
systematic edits (signs, 1000, Euro), relation
edits, macro edits, and so on
9
Example of a (small) process
Apply Data Validation Function
Population (a,q) (in progress)
Unit (a)
Unit (q)
Input population (a)
Apply Variable Derivation Function
Population (a,b,q) (in progress)
Population (a,q) (in progress)
Unit (a?q)
Unit (b)
Ready?
10
Examples of Statistical functions

• Data Validation Function
• Variable Derivation Function
• Error Localisation Function
• Error Correction Function
• Imputation Function
• Matching Function
• Estimation Function

11
(No Transcript)
12
Thank You
13
Glimpse of our framework
Statistical Strategy
Statistical Target
Set of Conceptual Attribute Variables
Statistical Method
Set of Operational Attribute Variables
Statistical Function
Design of Process Flow
Planning
Set of Physical Attribute Variables
Design of Statistical Process