摘要
加权最小绝对偏差法的搜索解法李显方（中南工业大学）ＳＥＡＲＣＨＭＥＴＨＯＤＯＦＷＥＩＧＨＴＥＤＬＥＡＳＴＡＢＳＯＬＵＴＥＤＥＶＩＡＴＩＯＮＳＭＥＴＨＯＤ￥ＬｉＸｉａｎ－ｆａｎｇ（ＣｅｎｔｒａｌＳｏｕｔｈＵｎｉｖｅｒｓｉｔｙｏｆＴｅｃｈｎｏｌｏｇｙ，Ｃ...
Abstract In this paper, multiple linear litting or regression (FR) problem is discussed by using weighted least absolute deviations method. It is proved that weighted absolute deviations function has the minimum, and there is a optimum FR hyperplane spaned by (m +1) observation points in (m + 1)-dimensional space. Then, Search procedure is introduced for estimating FR coefficients, some examples illustrate the effectiveness of this method.
出处
《计算数学》
CSCD
北大核心
1996年第2期183-188,共6页
Mathematica Numerica Sinica
