This study proposes a statistical framework for analysing vernacular architecture datasets characterised by sparse and heterogeneous distributions. Using an inventory of 1,212 buildings from eight villages in Portugal’s Montesinho Natural Park, the research evaluates the relationship between 23 architectural typologies and local settlements. Traditional Chi-square tests proved unreliable due to low expected frequencies, prompting the adoption of Standardised Residuals based on the Normal Distribution N(0,1). The method identifies statistically significant typological dependencies and local architectural identities while remaining accessible through standard spreadsheet software. Results demonstrate that rare typologies often provide stronger indicators of territorial identity than the most frequent building types.
