Related Experiment Videos
Markov regression models for time series: a quasi-likelihood approach
1Department of Biostatistics, Johns Hopkins University, Baltimore, Maryland 21205.
Biometrics
|December 1, 1988
Summary
This study introduces a quasi-likelihood (QL) approach for regression analysis in time series data. The method ensures accurate regression coefficients by correctly specifying the first conditional moment, even with complex observation-driven models.
Area of Science:
- Statistics
- Econometrics
- Biostatistics
Background:
- Regression analysis is crucial for understanding time series data.
- Existing methods may struggle with complex dependencies in time series.
- Observation-driven models offer a flexible framework for time series analysis.
Purpose of the Study:
- To present a quasi-likelihood (QL) approach for regression with time series data.
- To extend QL methods to observation-driven Markov models.
- To illustrate the application of QL for Poisson and gamma data.
Main Methods:
- Utilizing a quasi-likelihood (QL) framework.
- Applying it to observation-driven Markov models.
- Focusing on Poisson and gamma regression models for illustration.
Main Results:
- Large-sample properties of regression coefficients depend on the first conditional moment specification.
- The QL approach is robust for a class of time series models.
- Demonstrated applicability to count (Poisson) and continuous (gamma) data.
Conclusions:
- The quasi-likelihood approach provides a robust method for time series regression.
- Correct specification of the first conditional moment is key for reliable coefficient estimation.
- This method is applicable to a wide range of observation-driven models.