تجزیه و تحلیل مقیاس چند بعدی از سهام های مالی بر اساس ناپذیری کرونکر دلتا

Stock indices are usually regarded as the feedback on the related financial systems and environments, from which the relationship among financial areas can be detected. In this paper, a model is newly proposed, called multidimensional scaling based on Kronecker–delta dissimilarity (MDSK), which is a symbolic method to recognize signals with different properties. Experiments show that the length of the sequence does not affect the quality of the method, and MDSK is always a better choice than multidimensional scaling (MDS) methods with other alternative dissimilarity measurement we mentioned in this paper, particularly in the noisy environment. Our analysis reveals a clear clustering of eighteen indices from diverse stock markets, in which the geographical locations of indices in the same separated group are identical. Although both MDSK and MDS with other alternative dissimilarity measurement can all reach the clustering results, only MDSK separates the BVSP, an index from Brazil, as a single group instead of bracketing it with other indices from the North America. The results also imply that MDSK is more sensitive and superior than the MDS methods with other alternative dissimilarity measurement.