Accurate methods for incomplete data analysis using the mixed trunsored model and the decision tree.

本计画的目的是分析不完全资料的情形，其中资料是由非齐次资料所组成，借由估计模型参数与检验假设。为此，我们首先尝试将数据挖掘方案与不完全数据分析方法联合收割机相结合，证明了混合截尾模型在医学领域或可靠性领域等许多领域都具有适用性。IEEE Transaction的论文获得日本IEEE Reliability Society的奖项，表明了对该研究水平的高度评价，并给出了对杂乱数据分类有用的bump hunting方法的算法，以及在市场营销中的数值例子。本文在IEEE、IFORS、CNOSS等国际学术会议上作了大量的介绍，并结合电气绝缘应用的具体实例，提出了截断模型与决策树模型相结合的方法，达到了预期的目的。

The objective of this project is to analyze the incomplete data cases where the data are consist of inhomogeneous data by estimating the model parameters and testing the hypotheses. To do this, first we tried to combine the data mining scheme with the incomplete data analysis methods.We have shown that the mixed trunsored model has applicability to many fields such as the medical area or the reliability fields. The fact that the paper of IEEE Transaction was awarded by Japanese IEEE Reliability Society shows that the research level is highly evaluated.Also, the bump hunting method, which is useful to classify the messy data, has been shown with algorithms, and the numerical examples in marketing have been illustrated. We have had many presentations in IEEE, IFORS, and HICSS etc.The combined method of the trunsored model and the decision tree model has been developed with a concrete example in electrical insulation applications.Thus, we have reached the goal we posed successfully.