气候序列不连续点的小波分析和混合聚类分析

不论是单纯的小波分析还是单纯的非参数分析法都有其局限性.当样本量很大时,Fisher算法的计算量也是很大的,而且它不具有谱系结构,因此,为了保证谱系结构以及较快的计算速度,作者以小波变换和聚类分析相结合的方法来研究各类不连续点,给出了相应的算法和实验分析,并且实验结果给出了各类不连续点位置的比较精确的估计.

Discontinuous Wavelet Analysis and Mixed Cluster Analysis

We study the wavelet analysis and hybrid cluster analysis of discontinuous points, and explain that there are the shortcomings both in wavelet analysis and in nonparametric analysis. So, we detect each class discontinuities by means of wavelet analysis and hybrid cluster analysis. When the number of samples is great, Fisher algorithm has no hierarchical structures, and involves a mount of calculations. For maintaining hierarchical structure and gaining fast approach, we study each class of discontinuities by wavelet analysis and cluster anaysis. We present an algorithm and an experiment. From the result, we find that the method Performs well.