This research focuses on big data visualization that is based on dimensionality reduction methods. We propose a multi-level method for data clustering and visualization. It divides the whole data mining process into separate steps and applies particular dimensionality reduction method considering to analyzed data volume and type. The methods are selected according to their speed and accuracy. Therefore, we present a comparison of the selected methods according to these two criteria. Three groups of datasets containing different kind of data are used for methods evaluation. The factors that influence speed or accuracy are determined. The rank of investigated methods based on research results is presented in this paper.

No abstract

The amount of data in financial institutions is growing rapidly and the subject of "big data" has become an urgent trend. The "big data" phenomenon brings challenge to empower analytical methods for enhanced scope. At the same time the big data composed from various sources opens new possibilities to capitalize data research. The article investigates the anomalies in big data used by financial institutions. It proposes the model designed for exploring dynamics and detecting anomalous behavior of bank customers. The experimental screening on bank customers' big data shows significant time and necessary calculation steps reduction for detecting suspicious operations.

No abstract

No abstract