The main goal of this thesis is to examine time series and some clustering algorithms applied on them. The investigated time series are the stocks of the NYSE's S&P100 index. Both partitional and hierarchical types of classical clustering algorithms will be presented. Beyond the descriptions, the stocks had been clustered through the implementation of two algorithms (one partitional – one hierarchical). Beyond the above, the subject of this document is to represent and use some data processing techniques in order to achieve spectacular results with the clustering algorithms. Furthermore, some distance metric examples can be found, because one of the keystones of clustering is how we define distance.
After the algorithms have described and selected, there comes the testing phase with simulated and real data as well, in order to compare the functionality of the methods. Clustered Stock exchange data results can be compared with the grouping, based on natural attributes, hereby the similarity of the two grouping methods can be identified.