Decomposition and Visualization of High-Dimensional Data in a Two Dimensional Interface

Data visualization has a crucial role in understanding and processing voluminous data (i.e., Big Data) and subsequently has become more important with the coincidence of the exponential growth of data analysis need.The problem of high-dimensional data visualization in a data processing software interface cannot be entirely displayed, in consideration that once the data size exceeds two-dimension, it cannot be projected into a two-dimension interface. Furthermore, the rough analysis and evaluation of high-dimensional data become considerably ambiguous, thus, making a precise decision on that data cannot be achieved. In order to overcome this anomaly, resorting to data dimensionality reduction is a plausible solution.In this paper, the integration of Principal Component Analysis (PCA) combined with the Matrix by Block Decomposition (MBD) method(A.K.A block segmentation). According to the literature, the MBD method turned out quite efficient in data segmentation, wherein a huge data can be divided into regular blocks. By doing so, it becomes easier to access and visualize a given part of data. In order to further enhance the visualization understanding, K-means segmentation has been integrated in our proposed algorithm.In our study, we took into account other data dimensionality reduction techniques such as Linear Discriminant Analysis (LDA), Multi-Dimensional Scaling(MDS).