Efficient computation of hierarchical trends

To model a large database containing selling prices for houses, in which lo cal trends, general trends, and specific characteristics play a role, we de rived a new procedure to implement a state-space model for repeated measure ments. The original model is decomposed into two parts, which are treated d ifferently. The first part is ordinary least squares on data in deviation f rom means. This step provides a prior for coefficients to be used in the se cond step, which is a Kalman filter, providing estimates of the trends and the parameters. The procedure exploits and illustrates the Bayesian interpr etation of a Kalman filter.