罗吉斯蒂判别分析的稳健性

ROBUSTNESS OF LOGISTIC DISCRIMINANT ANALYSIS

罗吉斯蒂(Logistic)判别分析已广泛地应用于矿产资源评价、石油勘探、地理和环境科学等各个方面。它直接对两个总体的分布(密度)函数之比的函数形式作假设,而不考虑两个分布本身到底具有何种形式。本文试图从罗吉斯蒂判别分析数学模型和它在统计上的适用范围两个方面讨论其稳健性,从而说明它的实用价值。

Logistic discriminant analysis has been used widely in evaluation of mineral resources, oil exploration, geography, environmental sciences etc. In Logistic discriminant analysis, the attention is atracted to the functional form of the ratio of two distribution (density) functions which characterize two populations in study, not to two distributions themselves. In this paper, we attempt to discuss the robustness of Logistic discriminant analysis in terms of its mathematical model and statistical aPplicability, so that its value in practise can be identified further by geologists.