<p>The orientations of ancient tombs have attracted increasing scholarly attention, as they offer valuable insights into early social structures, cultural traditions, and the relationship between humans and their environment. However, the application of machine learning algorithms to the study of tomb orientation remains relatively underexplored. In this study, we employed a Gaussian mixture model to conduct a systematic analysis of the spatial and temporal evolution of Neolithic tomb orientations in Central China. We also examined the relationship between tomb orientation and both environmental factors and sociocultural dynamics. The findings suggest a deliberate and methodical approach to the planning and alignment of tombs during the Neolithic Age. Tomb orientations in each chronological phase displayed clear clustering patterns, reflecting a developmental trajectory from uniformity to diversity, and ultimately toward integration. While early angular measurement techniques appear to have emerged, they do not show evidence of sustained technical progression. Instead, different periods seem to have achieved similar levels of directional accuracy. The predominance of westward-facing tombs may be closely tied to both topographic features and the symbolic association with sunset. At the same time, cultural evolution and interregional exchange played essential roles in shaping the distinctive patterns of prehistoric tomb orientation. This research contributes not only to the understanding of ancient funerary practices but also demonstrates the potential of machine learning and artificial intelligence technologies in advancing archaeological analysis.</p>