基于可调权重距离的最小一乘回归方法

不同时期数据对因变量的影响权重不同,若不考虑数据的时间特性而构建回归模型,可能不一定得到最佳的回归模型.针对经济领域中数据时间跨度大,样本少以及可能存在异常点的特点,提出基于可调权重距离的最小一乘回归方法.建立了可调权重距离的权重系数确定方法,并给出基于MATLAB的模型求解方法.通过某船舶使用费用预测的应用,表明通过方法构建的模型具有更高的精度,值得借鉴.

Least Absolute Deviation Regression Base on Adjustable Weighting Distance

Data from the different periods have dissimilar influence to dependent variable.The best regression model may not be gained if not taking into account the influencing weighting of the different periods data.For large data time span,the small sample size and the possible existence of outliers,the least absolute deviations regression based on adjustable weighting distance is put forward.The method of determining the weighting coefficient based on adjustable weighting distance is made,and the solution method of the model based on Matlab is given.The result of the application instance on predicting the operation costs in a ship shows that the model founded by this method has better higher precision,which is worthy of referrence.