最优线性回归的计算方法

Algorithm for Optimum Linear Regression

本文指出利用常用的逐步回归方法可以计算出回归分析中常用的5种准则下的局部最优回归子集,而模拟结果显示,在大部分情形下,局部最优回归子集是相重合的.这就为逐步回归方法在应用上的重要性提供了科学依据.最后作者对现今著名的几个数字例子进行计算,其效果也是十分满意的.

The article shows how to find locally optimum regressor subsets using the well known stepwise regression algorithm for 5 commonly used criteria. Simulation result shows that in most cases the locally optimum subset coincide with the globally optimum subset. This finding provides good reason for the application of the stepwise regression method. Finally, the algorithm is applied to several well known real datasets and the results are satisfactory.