Dimensionality Reduction of Spatio-Temporal Data

Spatio-temporal data has become increasingly abundant due to the proliferation of sensors, mobile devices, satellites, and smart infrastructures. Such data, encompassing both spatial and temporal dimensions, is inherently high-dimensional, complex, and often redundant. Managing, analyzing, and extracting meaningful insights from spatio-temporal datasets poses significant computational and interpretational challenges. Dimensionality reduction techniques serve as powerful tools to mitigate these challenges by simplifying data without sacrificing critical information. This paper presents a comprehensive literature review on recent advances in dimensionality reduction methods applied to spatio-temporal data across various domains including climate modeling, remote sensing, video surveillance, transportation, and neuroscience. The review categorizes techniques into linear and nonlinear models, deep learning-based methods, and hybrid approaches, evaluating their suitability for different data characteristics and applications. Additionally, the paper highlights trends, identifies prevailing gaps, and discusses open research challenges such as preserving spatio-temporal correlation, scalability, and interpretability. This review aims to guide future research by mapping existing methods to application needs and motivating the development of robust, scalable, and context-aware dimensionality reduction frameworks

Data mining · Data reduction · Dimensionality reduction · Computer Science · Human Mobility and Location-Based Analysis · Mathematics · Remote-Sensing Image Classification · Time Series Analysis and Forecasting · Artificial Intelligence