A modular metadata-driven statistical production system The case of price index production system at...

17
A modular metadata- driven statistical production system The case of price index production system at Statistics Finland Pekka Mäkelä, Mika Sirviö

Transcript of A modular metadata-driven statistical production system The case of price index production system at...

Page 1: A modular metadata-driven statistical production system The case of price index production system at Statistics Finland Pekka Mäkelä, Mika Sirviö.

A modular metadata-driven statistical production systemThe case of price index

production system at Statistics Finland

Pekka Mäkelä, Mika Sirviö

Page 2: A modular metadata-driven statistical production system The case of price index production system at Statistics Finland Pekka Mäkelä, Mika Sirviö.

Background

Thoughts about a generic production and information system for price indices appeared around 2000 and preliminary planning started in 2004

The system has been designed, not only for calculation, but also for planning and administration of indices

During the project a need for a model of statistical production process emerged

Objectives Efficiency gains, reliability improvement Same data used for multiple purposes Reduction of data specific applications

Implications Common terminology, process, working methods

21.04.23 2

Page 3: A modular metadata-driven statistical production system The case of price index production system at Statistics Finland Pekka Mäkelä, Mika Sirviö.

Modularity in production of statistics

According to the principles of modularity the different phases of statistical production need to be standardized and independent of each other

The complexity of the system emerges from the interaction of the specialized (domain specific) modules

Modules only respond to inputs of a specific class and produce outputs of a specific class

The interaction of the modules requires standardization of interfaces between the modules

A strictly modular statistical production process can increase productivity but requires an advanced process management system

21.04.23 3

Page 4: A modular metadata-driven statistical production system The case of price index production system at Statistics Finland Pekka Mäkelä, Mika Sirviö.

Value chain of producing statistical information

21.04.23 4

The value chain of producing statistical information is a high level description of production process. It highlights the activities, that produce value to the customer, and their interdependencies.

The value chain offers two aspects to the production process: value creation and production costs.

Effectiveness of the value chain depends alike on the effectiveness and expediency of a single chain link and on the cooperation of the modules.

Critical factors in the functioning of the value chain are the interfaces between the modules and their standardisation.

Page 5: A modular metadata-driven statistical production system The case of price index production system at Statistics Finland Pekka Mäkelä, Mika Sirviö.

Determining product concept

Product concept is a high level definition of an information product including

main features of the product features and properties that define the product or product

group or differentiate it from other products customer-product interactions (e.g. potential use and

customers) Product concept describes the benefits of the product for the

customer and why the product is irreplaceable with other products, it doesn’t specify the implementation of the functionality and properties

21.04.23 5

Page 6: A modular metadata-driven statistical production system The case of price index production system at Statistics Finland Pekka Mäkelä, Mika Sirviö.

Creating statistical information

Creating statististical information for providing communication Organizing measured/empirical data and metadata to

standardized and validated data elements Summarizing data, estimating and validating the values of

population parameters based on data elements Creating specified presentation content of information

products analyzing of outputs and identification of core

elements/outputs description and interpretation of statistics (graphs, tables,

figures etc.)

21.04.23 6

Page 7: A modular metadata-driven statistical production system The case of price index production system at Statistics Finland Pekka Mäkelä, Mika Sirviö.

Process resource description model for production and information system for price indices Description model has four distinct elements

The values that guide the activities The activities consist of statistical production and various

supporting functions The values are complemented by the instruments that provide

practical means for the realisation of the values in the business process

The object of activity, input/output

21.04.23 7

Page 8: A modular metadata-driven statistical production system The case of price index production system at Statistics Finland Pekka Mäkelä, Mika Sirviö.

The key values General values

Fundamental principles of official statistics

European statistics code of practice

Human recource management strategy

Customer services policy Product concept and quality

specifications Production model (implemented

product concept) Data and metadata

The key instruments Technical instruments

Information and communication technology hardware and software

Communicative instruments Work culture, distributed

cognition Conceptual instruments

Terminological concept analysis Conceptual modelling Descriptive statistics, inferential

statistics

21.04.23 8

Page 9: A modular metadata-driven statistical production system The case of price index production system at Statistics Finland Pekka Mäkelä, Mika Sirviö.

Terminological concept analysis and modeling

Why terminological concept analysis is used for information system applications?

Terminological concept modeling produces valuable semantic information about concepts and concept relations.

Concept modeling produces easy to use and well applicable IT and concept systems, e.g. classifications in accordance with complete concept hierarchy.

21.04.23 9

Page 10: A modular metadata-driven statistical production system The case of price index production system at Statistics Finland Pekka Mäkelä, Mika Sirviö.

Computer aided conceptual modelling

Generic production and information system for price indices enables semi-automatic construction of concept systems, or ontologies

Some practical implementations Structuring of domain of statistics and creating classifications by

using delimiting characteristics as a subdivision criteria Characteristics are modelled by attribute - value pairs Common pool of characteristics for all concepts in system Generic product specification models using common pool of

characteristics Configuration of information products

21.04.23 10

Page 11: A modular metadata-driven statistical production system The case of price index production system at Statistics Finland Pekka Mäkelä, Mika Sirviö.

Creating classification, background

Concept is a unit of knowledge created by a unique combination of characteristics

Characteristics reflect shared properties of the objects belonging to the extension of a concept

Delimiting characteristic is abstract concept which consists of a set of concepts which are distinct and mutually exhaustive

Delimiting characteristics and their values can be used for subdividing a concept into several subconcepts

21.04.23 11

Page 12: A modular metadata-driven statistical production system The case of price index production system at Statistics Finland Pekka Mäkelä, Mika Sirviö.

Classification by delimiting characteristics

21.04.23 12

Domain: AdultsDelimiting characteristics: gender, parental relationship (PR)Values of gender: male, femaleValues of parental relationship: has PR, no PR Order of the delimiting characteristics:

gender, PR

Gender=male Gender=male, has PR (A) Gender=male, no PR (B)

Gender=female Gender=female, has PR

(C) Gender=female, no PR

(D)

Domain: AdultsDelimiting characteristics: gender, parental relationship (PR)Values of gender: male, femaleValues of parental relationship: has PR, no PR Order of the delimiting characteristics:

PR, gender

Has PR Has PR, gender=male (A) Has PR, gender=female

(C) No PR

No PR, gender=male (B) No PR, gender=female (D)

Page 13: A modular metadata-driven statistical production system The case of price index production system at Statistics Finland Pekka Mäkelä, Mika Sirviö.

Benefits of using delimiting characteristics as a subdivision criteria in creating classifications

Creating classification and to name created classes after predefined naming conventions are clearly separate and different tasks (modularity).

Every class in the hierarchy is connected with rich metadata which tell users the characteristics of concept and its relations to other concepts explicitly.

The framework of creating classifications presents and structures the information of the domain of statistics precisely.

Explicit sructuring of the domain of statistics enables automatic or semi-automatic processing of metadata by computers.

21.04.23 13

Page 14: A modular metadata-driven statistical production system The case of price index production system at Statistics Finland Pekka Mäkelä, Mika Sirviö.

Configurable information products

”A configurable product, or product family, is such that each product individual is adapted to the requirements of a particular customer order on the basis of a predefined configuration model, which describes the set of legal product variants (Sabin et al., 1998; Soininen, 2000).

Configurable products clearly separate between the process of designing a product family and the process of generating a product individual according to the product configuration model. This places configurable product in between massproducts and one-of-a-kind products by enabling customer specific adaptation without losing all the economical benefits of mass-products (Tiihonen et al., 1998).

21.04.23 14

Page 15: A modular metadata-driven statistical production system The case of price index production system at Statistics Finland Pekka Mäkelä, Mika Sirviö.

Examples of configuration in production and information system for price indices Indices with the same domain and nomenclature but which are

calculated differently: Different index formulas can be used as input in the system. Index formula in MathML format is used as a predefined component. Indices in accordance with new index formulas can be instantly

calculated. Each product has product specification:

The same product specification can be shared by many products. Product specifications are generated by characteristics modelled

by formal feature specifications, i.e. attribute-value pairs.

21.04.23 15

Page 16: A modular metadata-driven statistical production system The case of price index production system at Statistics Finland Pekka Mäkelä, Mika Sirviö.

21.04.23 16

Process management

Process control focuses on the requirements that the process has to fulfill (e.g. timetables, quality criteria, archiving, confidentiality)

Process analysis analyzes the key factors affecting the process in order to

guide and improve the process provides accurate and timely (possibly real time) information

on production of statistics for the production management and operative staff to enable fast reacting and to support process development

Page 17: A modular metadata-driven statistical production system The case of price index production system at Statistics Finland Pekka Mäkelä, Mika Sirviö.

Implementation timetable

2008 Index of real estate maintenance costs 2009 Price index of newly built dwellings 2010 Building cost index 2011 Deflator indices 2011 Index of producer prices of agricultural products 2012 Consumer price indices 2013 Producer price indices for services 2013 Producer price indices

21.04.23 17