It is well known that the regression coefficients are considered to be constant in the liner errors-in-variables (EV) model. In many applications, however, the coefficients may be varying with a covariate, such as time, temperature and so on. This paper gives such examples, and introduces how to estimate the coefficients when the coefficients are varying with a covariate in the structural linear EV model. Under the identification of cov-variance matrix of measurements error is known, we propose the adjust weighted LS estimators (AWLSE) for the estimated parameters of varying regression coefficients as well as the variance of model error. It is shown that the AWLSEs are strongly consistent and asymptotically normal under some mild conditions. Simulations illustrate our AWLSEs have good performance.
Cui Hengjian. , {{custom_author.name_en}}.
ADJUST WEIGHTED LS ESTIMATION FOR THE PARAMETER IN THE VARYING COEFFICIENTS LINEAR EV MODEL. Journal of Systems Science and Mathematical Sciences, 2007, 27(1): 82-92 https://doi.org/10.12341/jssms08738
PDF(417 KB)
