A New Dimensionality Reduction Approach Applied to the Big Data Visualization

Data visualization plays an important role in the analysis and processing of Big Data and this becomes more important with the explosive growth in the need to analyze and use data. The issue of Data visualization is manifested on that whenever it is voluminous and difficult to be interpreted, the extraction of relevant information becomes challenging. This fact impacts negatively for suitable decisions. Among the widely known problems involved in Big Data, we mention for instance duplication information. The latter has a drastic impact on its exploitation. These facts involve a high time consumption in terms of data analysis as well as its pre-processing. In this paper, we propose a twofold aim algorithm taking in consideration: (1) System (background processing) and (2) user (foreground processing). As regards the system, the algorithm permits to facilitate the comparison of data from high dimension to 2 dimension one (using Hu invariant moments approach). By doing so, we observed that the analysis complexity has been considerably reduced. Furthermore, the information redundancy has been eliminated. As concerns user side, our proposed approach renders the Data visualization within a screen more interactive and presentable.