【摘要】 本文提出了营销风险预警指标体系设计方法,给出了应用主成分分析法将原始风险预警指标体系浓缩为主成分指标体系的算法,设计了基于信息扩散理论的警度信号识别分类器。该分类器将主成分指标信息扩散到风险所有可能发生点上,其输入是浓缩后的主成分指标数据、而输出是期望的警度信号向量。基于该分类器,提出其虚警和误警概率的计算方法。由于主成分分析法的强浓缩性,因此本分类器具有处理高维预警指标体系的能力。更多还原

【Abstract】 This paper puts forward a method of selecting initial marketing early warning targets, gives an algorithm of condensing the initial targets into main components targets by the main components analysis method. A classifier is designed to estimate risk warning signal based on the information diffusion theory, the classifier diffuses information contained in the data of main components targets into all the points where risk may occur, the input of the classifier is the condensed main components targets, the output is the ex- pected warning signal vector. The possibility of error warning and false warning is calculated based on the classifier. Because of the high condensing ability the classifier possesses of processing targets with high dimensions.更多还原