定向判别分析新算法及其在沉积化学中的应用

New arithmetic method of directional discriminant analysis and its application on sedimentary chemistry

介绍了处理多元有序数据的定向判别分析新方法原理、建模流程、应用流程及其在沉积化学中的应用实例。这种判别分析将分类建模与判别归类分开,求解与专业知识结合。新方法用多组或逐步判别分析对多元有序数据建模,应用时根据应用领域的知识对样本归属作初步定向,然后选择模型的相关局部进行判别归类,从而实现有序判别。这种方法用于解决由于时间序列多元数据周期性造成的样本分类颠倒问题。在塔里木盆地沉积岩时间序列化学数据的应用实例中,解决了石油井下地层预测和归类问题。

The principle, modeling flow chart, applying flow chart and the application on sedimentary chemistry of a new arithmetic method called Directional Discriminant Analysis （DDA） are introduced to process multivariate time series data. In DDA, the class modeling and data discriminating are separated, and the solution is conbined with specialty knowledge. The model to multivariate time series data is built by multiple or stepwise discriminant analysis. While applying the model, the initial estimation of the samples＇ classification should be given according to the knowledge in the problem-corresponding field. Then the program selects appropriate parts of the model to discriminate the classes of the data. Thus, sequential discrimination is carried out. In this way, the upside down problem of sample classification caused by the periodicity of multivariate time series data could be solved. While applying the method to the time series data of sedimentary rock in Talimu basin, the prediction and classification of layer in petroleum well have got solution.