Transcript MSIS 2007 - Data Warehousing

Data Warehousing
at STC
MSIS 2007
Geneva, May 8-10, 2007
Karen Doherty
Director General Informatics Branch
Statistics Canada
Table of Contents
•
•
•
•
Canadian Systems of National Accounts
SNA Warehouse
Data Warehouse Framework
Lessons Learned
2
System of
National Accounts
• The SNA provides a conceptually integrated
framework of statistics and analysis for
studying the state and behaviour of the
Canadian economy
• The accounts are centered on the measurement
of activities associated with production of
goods and services, the sales of goods and
services in final markets, the supporting
financial transactions, and the resulting wealth
positions
3
System of
National Accounts
• Input-Output Division (now the Industry Accounts
Division)
– produces the output, input, and final demand tables for each
provincial and territorial jurisdiction
• the tables are linked through an interprovincial flows table that
shows imports and exports between jurisdictions
– covers all economic activities (persons, businesses, government
and non-profit organizations, and external entities that generate
imports or exports (interprovincially or internationally)
• The I-O tables represent the most detailed accounting of
the Canadian economy available and thus serve as
benchmarks to the Canadian System of National
Accounts
4
I-O Re-engineering
• Project Impetus
– aging production systems and work processes
– no tools for data verification and table balancing imposed a heavy burden
on staff
– lack of integration and standardization of processes and procedures
impeded the division’s ability to handle the growing amount of input data
• Project Objectives
– maximize knowledge retention and reuse through the introduction of
software to specify derivation and balancing methodologies
– maximize operational integration potential of the various divisions of the
SNA through the introduction of a data management system to integrate
data and meta-data
– facilitate data reconciliation between I-O and other SNA divisions
– maximize analytical potential through the introduction of main stream
analytical tools to detect problems in the I-O Tables and source data
5
Solution
• A data warehouse with three components:
– user-supplied micro and aggregate data to facilitate
data confrontation
– aggregate (macro) data to support data reconciliation
with the GDP outputs from system divisions
– tools for analysts
• standard statistical functions
• tools to calculate industry and commodity specific ratios
6
System Capabilities
• Analysts can:
– compare data from different sources, ensuring consistency of
estimates by making dissimilar classifications comparable
– reconcile information across divisions in the SNA and make
effective decisions during the annual production cycle
– compare statistics in terms of ratios, proportions, growth rates,
by region and in chronological series
– review the metadata and information on how the data was
established, concepts and definitions, classifications and
concordances and best practices with respect to processing or
analytical procedures
– create reports which are automatically updated whenever they
are opened
– perform graphics-based analysis
7
Results
• Phase 1 – I-O Division
– standardization of the analytical process
– enhanced data coherency
– standardized and normalized analytical procedures
which allows the division to operate with less
experience staff
– more transparent, repeatable and efficient analysis of
data
• Phase 2 – SNA
– the success in I-O led to an SNA-wide warehouse
8
I-O Data Warehouse
Architecture
9
Data Warehouse
Framework
10
Technology
• Microsoft Data Warehouse Framework for
SQLServer 2000
– currently working on the SQL Server 2005
version
– includes the MS Enterprise Manager, Data
Transformation Services (DTS) and Analysis
Services
– fully integrated with Microsoft Excel XP
11
Technology
• Cubes
– based on the OLAP standard
• Data Transformations
– any Extract, Load, Transform (ETL) product can be
used but the team has standardized on the Microsoft
product
• API
– XML for Analysis standard
– uses Microsoft’s MDX query language
12
Technology
• Reporting and End-user Tools
– EzWeb OLAP Report Browser and EzWeb OLAP Report Designer
• developed by the Data Warehouse Web team at STC
• provides a web like interface that conforms to the Government of
Canada’s Look and Feel Standard
• Microsoft Office Web Components (OWC) Pivot Table provides OLAP
functionality
• can navigate from one OLAP report to another
• Data Marts provide users with a customized subset of data and reports
– Excel XP
– Business Intelligence tools, Data Mining tools, etc.
• STCWiki (in pilot mode)
– implemented using MediaWiki (product used by Wikipedia)
– two-way communications with STC’s Integrated Metadata database
13
Lessons Learned
• Technical
– standardized framework greatly reduces development
costs
– loosely connected customized data marts are more
effective:
• partitions the effort involved in harmonization and the
management of security and access rights
• allows users to customize their personal portal to list only those
sources which are of business interest to them
– appropriate for any type of data (operational, data
processing, analysis of published data, etc.)
14
Lessons Learned
• Business
– the real challenge lies with how data should be
processed, analysed and classified (data harmonization)
– value gained by harmonization usually results in
modified working procedures
– start by having good analytical tools to allow business
units to detect problems and improve and adapt the
methods used to ensure that the data is of high quality
15