Data holds a crucial role in the standard system of input, processing, and output, particularly representing the input. With high-quality data, there is an assurance of a smooth process and an effective outcome. However, the issue of privacy in data acquisition has become a significant concern in recent years, forming the core of numerous research studies. This paper introduces a novel approach to generating synthetic data that retains the same statistical structure as the original, based on the geometrical representation of each data point. This newly generated synthetic data can act as a substitute for the original, ensuring the integration of privacy into any application utilizing the data. Experimental results on two datasets of varying sizes demonstrate that the proposed method can effectively produce synthetic data with the same statistical structure and a comparable level of accuracy for predictive tasks as the original data