zbMATH — the first resource for mathematics

The truncated least squares method for a class of autoregressive models. (Romanian) Zbl 1084.62515
Summary: We determine the parameters $$a_j$$ and $$b_j$$ of a Kremer type autoregressive model $X_{ij}=a_i+(b_j+r_{ij})X_{i,j}+e_{ij},\;i=\overline{1,n},\;j=\overline{2,n},$ using the truncated least squares method. We reduce the statistical problem to minimizing a concave function over a polytope, for wich we present an iterative procedure for the determination of a local optimum. In the end, adaptation of Tuy’s cutting plane method is used for the construction of the global optimum of our problem.
MSC:
 62M10 Time series, auto-correlation, regression, etc. in statistics (GARCH) 65C60 Computational problems in statistics (MSC2010) 65K10 Numerical optimization and variational techniques
Full Text: