Recursive algorithms for nonstationary regression

Recurrent algorithms are considered for computing the estimates in a nonstationary regression problem. Their convergence is proved and the rate of convergence studied. An asymptotically efficient algorithm is obtained.