🌐       https://exaly.com/author/8213659/rahim-mahmoudvand

Cities and regions generate vast amounts of data across multiple domains such as healthcare infrastructure, socioeconomic conditions, demographic indicators, and environmental factors. These variables are often analyzed using conventional statistical methods that may overlook underlying spatial structures and patterns.

Singular Spectrum Analysis (SSA) is a powerful signal extraction technique traditionally applied to time series data. However, SSA does not inherently require time-based ordering; but it can analyze any sequentially ordered data. This opens the possibility of applying SSA to spatially referenced data by creating meaningful geographic sequences.

This paper introduces a methodology for spatial signal extraction using SSA. The approach involves ordering geographic units along different spatial axes, latitude, longitude, and rotated diagonal directions, to create sequences that SSA can decompose into trend, oscillatory components, and noise. A novel rotation optimization procedure uses linear regression of coordinates to identify the natural geographic axis of the study region, maximizing signal extraction along the dominant spatial orientation.