NbbTools >
Documentation Applications
Generalities
The
"NbbTsAnalyser" application is an advanced graphical tool that
performs the following algorithms:
The model-based time series decompositions involved in Tramo-Seats, in the
modified Hodrick-Prescott, in the basic structural models and in the generalized
airline model are presented following the (semi-)infinite Wiener-Kolmogorov
approach developed in Seats. Estimates rely on the Burman-Wilson algorithm or on
the Kalman smoother.
Most of the algorithms share other important features, like the calendar
effects correction, the residuals analysis or the likelihood function display.
Back
to Top
Quick User's guide
Data workspace
The data processed by the application are stored in a set of time series
collections (called a workspace). The workspace
is visualized by the tree on the top left panel of the main window.
Usual operations, like
saving, loading, ... are available through the file menu items.
Back
to Top
Importing the data
Time series can be imported in the workspace using copy-paste operations
("edit →
paste in
workspace" menu item) or using drag and drop, the target zone being the
workspace tree window.
Different
formats are supported. The most important ones are the Excel format (Excel 2003
or higher) and the Text format.
Importing
data from Excel
- The series in Excel can be vertically or horizontally oriented
but the first column to the left of the observations or the first row up
to the observations depending of the orientation must be real Excel
dates. This point also applies to yearly series.
- You must always select the dates together with minimum one column or one
row of observations.
Examples of good selections
:
You can select
only a part of a table without the titles. The application will automatically
give names for each series (Series 1, Series 2 ....).
If you want all
the table observations and if there are titles joined to the series, you can
select them together with the observations. These names will be displayed as
titles in the application (a series without name remains without name in the
application).
Back
to Top
Importing
series in the text format
-The data must
be structured very similar to the vertical orientation used for Excel.
-The text must
be TAB delimited for the different columns.
-The dates must
be in the "Short date"
format as defined in your "Regional
Options" (Start - Settings - control Panel - Regional and Language
Options).
Back
to Top
Data analysis
The different analyses
of a time series are organized around a "TsAnalyser" window. Such a
window is created by the "File →
New → Analysis" menu sequence.
The time series that
must be analyzed is imported in the TsAnalyser window through a double click
from the workspace tree or by drag and drop. Direct imports from external
sources are also possible. If several TsAnalyser windows are open, the double
click selection only affects the default window (usually the last one).
Actual analyses can then be executed either from the main menu (TsAnalysis
→
Tramo/...) or from the local menu.
Several analyses can be
performed on the same series. For instance, you can create one (or several)
Tramo(s), Basic Structural Model(s)
and Generalized Airline Model(s) from only one TsAnalysis window. If the series
of that window is changed, all the linked analyses are automatically updated. By
closing the parent TsAnalysis window, all the related algorithms views are
destroyed. However, an algorithm view can be safely closed and re-created later
through the local menu (Show command),
without losing the current specifications (see below).
Back
to Top
Algorithms specifications
Each algorithm has its own specifications that can be modified through a
dialog box. It can be called through the main menu (Tramo/... → Specifications...). Changes in the specifications are immediately
reflected on the output, after that the "Apply" button of the
specifications dialog box has been pressed.
Other remarks
Many parts of the output provide local menus. As usually, they can be called by
means of the right mouse button. Moreover, most of the output can be transferred
to other software by copy/paste or drag and drop operations.
Back
to Top
Description of the algorithms
It is out of the scope
of this quick user's guide to describe the different algorithms. However, you
can find below some references for each of them.
Tramo-Seats.
Gomez V. and Maravall
A. (2001), "Automatic Modeling Methods for Univariate Series" Ch. 7 in
D. Peña, G.C. Tiao, and R.S. Tsay, eds., in "A Course in Time Series
Analysis", New York: J. Wiley and Sons , 170-201.
Gomez V. and Maravall A.
(2001), "Seasonal Adjustment and Signal Extraction in Economic Time
Series", Ch. 8 in D. Peña, G.C. Tiao and R.S. Tsay, eds., "A Course
in Time Series Analysis",
New York: J. Wiley and Sons, 202-246
Maravall A. (2002),
"Brief Description of the Tramo-Seats Methodology", Proceedings of the
3rf International Symposium on Frontiers of Time Series Modeling, The Institute
of Statistical Mathematics, Tokyo.
Many other papers are available on the site of the Bank of Spain.
(http://www.bde.es/servicio/software/papers.htm)
Modified
Hodrick-Prescott
Kaiser R. and Maravall A. (2002), “A Complete Model-Based Interpretation of
the Hodrick-Prescott Filter: Spuriousness Reconsidered“, Working Paper 0208,
Servicio de Estudios, Banco de España.
Basic
Structural Model
Durbin J. and Koopman S.J. (2001), "Time Series Analysis by State Space
Methods". Oxford University Press.
Harvey, A.C. (1989), "Forecasting, Structural Time Series Models and the
Kalman Filter", Cambridge University Press.
X-11
Ladiray D. and Quenneville
B. (1999), "Comprendre la méthode X11"
(http://www.census.gov/srd/www/x11doc_abs.htm)
Generalized
Airline Model
John A. D. Aston, David F. Findley, Kellie C. Wills, and Donald E. K. Martin
(2004), "Generalizations of the Box-Jenkins Airline Model with
Frequency-Specific Seasonal Coefficients and a Generalizaton of Akaike’s
MAIC", presented at 2004 NBER/NSF Time Series Conference
(http://www.census.gov/srd/www/sapaper.html).
Back
to Top