An Efficient Numerical Method for Estimating Regression Parameters
In the study of linear regression parameter estimation,traditional gradient descent and Newton's method are often limited by local convergence issues and are highly sensitive to the choice of initial values and learning rates.To overcome these limitations,a regression parameter estimation algorithm based on central difference is pro-posed,which uses central differences to approximate the partial derivatives of the loss function,effectively reducing the model's dependence on learning rate adjustment and initial value selection.The experimental results of the line-ar regression model of children's urine creatinine content and age show that the algorithm has significant advantages in terms of computational efficiency and initial value sensitivity,which can greatly improve convergence speed and accuracy.The goodness of fit and mean square error of the algorithm are very close to the least squares method,demonstrating excellent robustness and wide applicability.
linear regressionparameter estimationcentral differenceNewton's method
国家科技期刊平台
NSTL
万方数据
