Contents
Course Description
Course Programme
Request an Enrolment Form Now
Terms and Conditions
Timberlake Consultants Ltd., the UK distributor of STATA, invites you to attend a 2-day Advanced Course. This course will cover the theory and application of state space models to time series analysis and forecasting. It will also cover the theory and application of dynamic factor analysis to problems in the social, financial, econometric and policy sciences. Analytical examples will use Stata, space and dfactor, the well-known software package developed by StataCorp (USA). Data sets will come from these subject areas, but students may bring their own datasets.
Who should attend: Persons interested in the theory and application of state space models to solving problems of signal extraction, interpolation, forecasting and control in time series analysis and forecasting as well as those interested in applying dynamic factor analysis to solving problems in the social, financial, econometric and policy sciences. Example data sets will come from these areas.
Mathematical Background Required
Prerequisites: Basic Stata, basic linear algebra, some multivariate statistical analysis, and a basic course in time series analysis.
Recommended: Basic time series analysis with Stata
The Principal Lecturer: Robert A. Yaffee, Ph.D., a research professor at New York University and a senior research scientist/statistician on a U.S. National Science Foundation grant, served as a senior research/statistical consultant at the Academic Computing Services of the New York University Information Technology Services from 1989 until spring 2004. Dr. Yaffee is author of a forthcoming book entitled An Introduction to Forecasting Time Series using Stata (expected publication date winter 2009-2010), and an author of a recent textbook entitled An Introduction to Time Series Analysis and Forecasting with Applications of SAS and SPSS (Academic Press, 2000). Yaffee has written articles on the design and planning of statistical analysis, logistic regression analysis, along with a number of articles on the psychosocial aspects of pathological gambling. He has lectured on the research methods in empirical research, theory and programming of structural equation models, event history analysis, complex sampling, categorical data analysis, time series analysis, and quantitative epidemiological analysis.
From 1995 through 2000, he held the position of research scientist/statistician at Downstate Medical Center, working under a National Institute of Mental Health grant to study depression and anxiety on the part of immigrant groups within Brooklyn. Before joining New York University, he served as an associate research scientist at the Columbia University School of Public Health on a National Institute for Drug Abuse grant. From 1986 through 1990, he served as a member of the editorial board of the Journal of Gambling Behavior and from 1990 to 2004; he has served on the editorial board of the Journal of Gambling Studies.
The course fees are: The cost includes course materials (handbook, all models, templates and add-ins), as well as complimentary lunch, refreshments and the use of computers. The number of delegates is restricted. Please register early to guarantee your place.
Further instructions will be sent with the joining instructions. If you need assistance in locating hotel accommodation in the area, request the help of our Training Department.
Cost - The course fees are
|
Prices
|
2 day course only |
| Non-Academic |
$1500
|
| Academic |
$1000
|
More discounts available:
20% discount if you register by January 29th, 2010
The cost includes course materials, lunch, refreshments and the use of computers. The number of delegates is restricted. Please register early to guarantee your place. If you need assistance in locating hotel accommodation in the area, request the help of our Training Department.
Agenda
(subject to minor changes)
DAY 1
| AM | Registration |
| Introduction to Dynamic Factor Analysis | |
| Data preparation | |
|
Principal Components Analysis |
|
| Assumptions: stationarity and completeness | |
| Construction of the correlation matrix | |
|
Type of correlation to use |
|
| Canonical decomposition of the correlation matrix | |
|
eigenvalues and eigenvectors |
|
|
Component extraction |
|
| Number of components to extract | |
| Rotation of components | |
|
Orthogonal |
|
| Oblique | |
| Component score generation | |
| Examples with portfolio asset returns | |
| Application of component scores to other models | |
| PM | Dynamic factor analysis |
| Static vs. dynamic factor analysis | |
|
Assumptions: Stationarity, normality, and linearity of dynamic factors. |
|
|
Data preparation |
|
|
Construction of the covariance matrix |
|
| Canonical decomposition of the covariance matrix | |
| Communalities in principle diagonal | |
| Factor Extraction Techniques | |
| Single common factor models |
|
|
Statistical tests: BIC, AIC |
|
|
Overfactoring (Heywood cases) |
|
|
Multiple factor analysis of portfolios and macroeconomic series |
|
|
Stata dfactor (Dynamic common factor analysis) |
|
| Seemingly unrelated regression models with dfactor | |
|
Single common factor models with dfactor |
|
|
Static factor models |
|
|
Dynamic factor models with dfactor |
|
|
Estimation |
|
|
dfactor postestimation: Reliability analysis |
|
| forecasting with dynamic factors |
DAY 2
| Early AM | State Space Models |
|
The Kalman filter |
|
|
Classical Diffuse initialization |
|
| The Kalman filter smoother | |
|
State and Disturbance smoothing Forecasting |
|
| The local level model | |
| The local linear trend model | |
| Late AM |
Seasonality |
| Cyclicity | |
|
Interventions: Additive outliers, outlier patches, level shifts, ramp effects, periodic pulses, and segmented trends. |
|
|
Component extraction |
|
| Exogenous covariates | |
| Splines | |
|
The general dynamic state space model |
|
| Early PM | Programming Stata sspace |
| Data preparation | |
| Principal Components Analysis | |
|
Cointegration and factor-augmented error correction models |
|
|
Stata sspace |
|
|
Model identification and constraints |
|
|
Model Building |
|
|
Model fitting |
|
| Late PM | sspace postestimation |
| Forecasting state vectors | |
| Forecast evaluation of factors | |
| Computer hands-on practice |
Registration closes 5 calendar days prior to the start of the course.
Cancellations:
Timberlake Consultants BusinessTerms and Conditions