Computing tasks associated with Time Series Processing

1
Computing tasks associated with Time Series
Processing

• Extract from a presentation
• by Fortier and Quenneville
• Statistics Canada -TSRAC
• BSMD Seminar -- October 11th, 2006

2
Time series analysis package

• Computing tasks associated with a typical
  seasonal adjustment process at STC
• New Time Series Generalised System
• Switching to X-12-ARIMA.
• Updating our benchmarking and reconciliation
  methodologies and softwares.

3
Modules of the Prototype

• Simplified diagram of the prototype
• Mostly SAS-based
• Programmed by computer experts in system
  development
• For production purpose

pre
X12
Bench
post
Rake
Graphs Tools
Trend
4
Modules of the Prototype

• The SA is done through X-12-ARIMA
• Pre and Post steps mainly to validate inputs and
  verify outputs.
• Some pre/post adjustment capabilities as
  safeguard

• With the Benchmarking module
• SA data can be benchmarked to annual totals (from
  original data or external source).
• Benchmarked could also be applied to original
  data prior to SA.

pre
X12
Bench
Bench
post
Rake
Graphs Tools
Trend
5
Modules of the Prototype

• With the Raking module
• aggregation constraints are restored (direct SA)
• indirect SA ( simple sum)
• revision strategy implemented

• Because of possible raking and revision strategy,
  the trend can not be computed with the initial SA
• Trend is calculated in a later step on published
  data (presumably in X12)

pre
X12
Bench
Bench
post
Rake
Graphs Tools
Trend
6
Modules of the Prototype
pre
X12
Bench
Bench
post

• The Graphs and Analytical Tools helps with the
  preparation of the analysis that is published
  with the release of SA data.
• Growth rate analysis

Variance
Rake

• The Variance module with results from Chens
  research (X11) and considerations for raking and
  benchmarking.

Graphs Tools
Trend
7
Expert System

• An Expert System is also in the plans to help us
  select the SA options for large sets of data.

pre
X12
Bench
Bench
post
Variance
Rake
Graphs Tools
Trend
8
Benchmarking

• Forillon project
• A SAS procedure developed at Statistics Canada
• User-friendly interface (Developped using SAS
  Enterprise Guide
• Provide input information to the procedure
• Mainly for training and demonstration
• Easy access to results in graphical format
• Available graphs for Benchmarking
• Original scale
• Adjustment scale ( Benchmarked to Indicator
  ratio)
• Growth rates

9
(No Transcript)
10

• For more information please contact

• Pour plus dinformation veuillez contacter

Susie.Fortier_at_Statcan.ca Benoit.Quenneville_at_Statca
n.ca