![]() | Only 14 pages are availabe for public view |
Abstract The prevalent issue of the thesis is to get a new time series classification method to classify (or allocate) an observed time series {Xt, t = 1, 2, …, N} to one of the k populations (or categories) 1, 2,..., k with minimal error margin. Time series classification has drawn a lot of attention in literature. Thus, a method of classification using frequency domain is implemented to evaluate the linear and quadratic functions. Further, the classification by motifs, using SAX approximation, is used as well to give better accuracy than adopting other methods of classification; concerning calculations, a number of packages and programming languages (R 3.5.2, Python, Matlab 2018, Minitab 17) are used. The scope of the thesis is then roughly as follows. |