The present work aims to quantitatively explore and understand the relationship between mobility types (nautical versus pedestrian), specific technological traits and shared technological knowledge in pedestrian hunter-gatherer and nautical hunter-fisher-gatherer societies from the southernmost portion of South America. To that end, advanced statistical learning techniques are used: state-of-the-art classification algorithms and variable importance analyses. Results show a strong relationship between technological knowledge, traits and mobility types. Occupations can be accurately classified into nautical and pedestrian due to the existence of a non-trivial pattern between mobility and a relatively small fraction of variables from some specific technological categories. Cases where the best-fitted classification algorithm fails to generalize are found significantly interesting. These instances can unveil lack of information, not enough entries in the training set, singular features or ambiguity, the latter case being a possible indicator of the interaction between nautical and pedestrian societies.
Keywords: hunter–gatherer; mobility; random forest; shared technology; statistical learning.