THE SEQUENCIAL REGION SEARCH METHOD FOR LEAST SQUARES ESTIMATION OF NONLINEAR REGRESSION

In this paper a new method——Sequencial Region Search method （SRS method） is given for obtaining least square estimates of nonlinear regressions. A number of examples show that in comparison with the iterative procedures the SRS method is less sensitive to the initial values, but more effective, and the algorithm of SRS method is simpler. Also, a large amount of simulating examples is worked out and the results are satisfactory.