Date post: | 27-Mar-2015 |
Category: |
Documents |
Upload: | kaitlyn-weber |
View: | 213 times |
Download: | 0 times |
Data Model and Matrix Schemas
XI European Banking SupervisorsXBRL Workshop in Vienna19th November 2009
Katrin SchmehlMember of CEBS XBRL Network
Agenda
Multidimensional terminology
Appliance on FINREP framework
Overview on matrix schema
Multidimensionality of data
Product group
Books News papers
Magazines Maps
Co
un
try
France 500 452 124 35
Spain 852 634 236 85
Austria 632 234 963 45
Belgium 459 325 456 96
Dimensional view on gross revenue
Multidimensionality of data - Measures
measures
pro
du
ct g
rou
p
country
Gross revenue
Books
France
500
Measures: data mostly in numerical dimensions
…
500
Multidimensionality of data - Dimensions
Dimensions: views or objects for which data should be reported,
specific breakdown of a measure, ie. “by country”
measures
pro
du
ct g
rou
p
country
Books
France
…
500Gross revenue
Other dimensions
country of origin
buying country
Multidimensionality of data - Domain
Domain: a set of items that have a specific coherence
Books
France
…
country
France
Spain
Austria
Belgium
Germany
Italy
DomainDomain
Sub domainSub domain
500Gross revenue
Multidimensionality of data – Domain member
Domain member: a specific member of a domain
Books
France
…
country
France
Spain
Austria
Belgium
Germany
Italy
DomainDomain
Domain memberDomain member
500Gross revenue
Multidimensionality of data – Data point
Product group
Books News papers
Magazines Maps
Co
un
try
France 500 452 124 35
Spain 852 634 236 85
Austria 632 234 963 45
Belgium 459 325 456 96
Dimensional view on gross revenue
634
Multidimensionality of data – Data point
gross revenue
product group
country
News papers
Spain
Multidimensionality of data – data base
Facts
ID_Product groupID_CountryID_PeriodGross revenue
CountryIDname
Product groupIDname
PeriodIDyearmonthday
Dimensions have the character of keys for multidimensional data
Multidimensionality of data - Hypercube
Hypercube: binds dimensions together in order to build a Cartesian product
Measure
Hypercube Dimension1 Domain a b
cd
Dimension2 Domain A
B C
D
gross revenue
product group
country
News papers
Spain
measure
pro
du
ct g
rou
p
country
Multidimensionality of data - sections
Product group
Books News papers
Magazines Maps
Science fiction
Romance Thriller Total
Co
un
try
France 100 200 200 500 452 124 35
Spain 152 300 400 852 634 236 85
Austria 200 232 200 632 234 963 45
Belgium 240 100 119 459 325 456 96
Dimensional view on gross revenue
by November 2009
Multidimensionality of data – definition of sections
Dimensional views on gross revenue Cartesian product
16 Combinations of one member of product group with one member of country
12 Combinations of one member of product group with one member of country and one member of book category
Multidimensionality of data – definition of sections
Cartesian product defined by a hypercube
16 Combinations of one member of product group with one member of country
12 Combinations of one member of product group with one member of country and one member of book category
Same section (same extended link role)
All
All
Set of combinations
Set of combinations
The intersection of both sets is null no combination is allowed.
VALIDATION!
Multidimensionality of data – definition of sections
Cartesian product defined by a hypercube
16 Combinations of one member or product group with one member of country
12 Combinations of one member or product group with one member of country and one member of book category
Different sections (different extended link roles)
All
All
Set of combinations
Set of combinations
The union of both sets every combination is allowed.
VALIDATION!
Agenda
Multidimensional terminology
Appliance on FINREP framework
Overview on matrix schema
FINREP Data model – measures and dimensions
FINREP Data model – measures and dimensions
dimension
dimension
dimension
dimension
measuresmeasures
dimension
dimension
FINREP Data model – definition of sections
Section 1Section 1
Section 2Section 2
Section 3Section 3
FINREP Data model – definition of sections
Section 4Section 4
FINREP Data model – Common elements
Data points are repeated
Identification should be the same !
same primary
same dimensions
same dimension members
CONSISTENCY
STABILITY
FINREP Data model – Movement tables
Data refers
to an instant in time (opening balance and closing balance)
to a duration in time (variations in time)
Possible solutions
Two primaries (two dimensions)
Primaries in the columns
Primaries in the rows
CONSISTENCY
STABILITY
Agenda
Multidimensional terminology
Appliance on FINREP framework
Overview on matrix schema
Matrix schema – What is a matrix schema?
Representation of the multidimensional structure of XBRL
taxonomies in Microsoft Excel
Generated by a reverse-engineering process of the Bank of
Italy
Initially designed for reviewing COREP and FINREP
taxonomies
Does XBRL match the business requirements?
Matrix schemas are used since several years in the CEBS
XBRL
Network
XBRL taxonomyXBRL taxonomy Matrix schemaMatrix schema
?
Matrix schema – How is it now used?
FINREP tablesFINREP tables Data classificationData classification
XBRL taxonomyXBRL taxonomy Matrix schemaMatrix schema
?
Matrix schema - How should it be read?
measures reference to sub domains
Matrix schema – The domain / sub domain list
Matrix schema – The advantages
Easy to read and understand (in comparison to XBRL)
Very compact format
Eases the review process for the data classification
Shows the matching between business requirements and the
data model to be implemented
Tool for quality assurance
Prospects: Consolidated matrix schema approach