多状态系统的LAD决策树的建树方法研究

LAD( Logistic Data Analysis Tree)是一种逻辑数据分析技术 ,它将布尔代数和优化分析方法引入到了判别分析领域 ,提出了一种布尔变量集合的变量筛选和建模方法 ,并可以对冗余模式进行可视化识别与删除 .但目前的 LAD技术还仅限于二状态 ,而且算法复杂 .本文将 LAD决策树推广到了多状态情形 ,以三状态下建立 LAD决策树为例 ,提出了不可分辨度的定义 ,并以其下降最大作为寻找最优决策树的依据 .说明多状态下建立 LAD决策树的计算方法及重要的算法步骤 .最后 ,本文以鄱阳湖地区洪涝灾害影响研究为案例 ,采用 LAD决策树方法对其进行判别分析 .

Research on Modeling Method of LAD Decision Tree in Multi-state Condition

In this paper, Logistic Data Analysis (LAD) decision tree is generalized to multi state discrimination analysis. The definition of undistinguished degree is proposed and maximum decrease of undistinguished degree is suggested as the principle to search optimum LAD decision tree. In order to show the calculation steps of LAD decision tree in multi state condition, the modeling method of three state condition is put forward as an example. Then, the method of LAD decision tree is adopted to discriminate the pattern of flood disaster in Poyang lake area.