“It’s all just barley and figs!” Identifying patterns of plant waste accumulation in House 169, Elephantine Island, Egypt (1750–1650 bc) using machine learning

A study of archaeobotanical samples recovered from excavations by the German Archaeological Institute (2013–2019) in late Middle Kingdom (1750–1650 bc) house 169 on Elephantine Island (Egypt) yielded an assemblage of exceptionally well-preserved desiccated and charred plant macro-remains. In order to determine patterns within the assemblage of 208,493 plant items from 123 archaeological contexts, we applied a novel approach to multivariate statistical analyses, making use of machine learning methods. A combination of cluster analysis and evolutionary decision trees out-performed traditional correspondence analysis and principal components analysis techniques, allowing for the discovery that whilst the assemblage was generally homogeneous, it was possible to identify certain types of deposits as being locations of plant “waste” disposal. The analysis also allowed identification of possible storage of barley and flax processing in one room, and some fireplaces in which the inhabitants chose to burn specific selected materials.