Multiple approaches to predicting flake mass

Predicting original flake mass is a major goal of lithic analysis. allows for researchers to make estimations remaining mass, lost and other features. All these measures relate the organization technology by past societies. The present work tests three different models predict log mass: multiple linear regression, random forest artificial neural networks (ANN). Estimations were performed using features flakes from an experimental assemblage. This assemblage was obtained expansion previous dataset through inclusion bigger flakes, allowing analysis account effects sample size value distribution. Correlation results show large/strong relation between predictions real outcome (r2 = 0.78 in best case). Comparison affords insights into variable importance predicting mass. Results that (for dataset) regression still stands as method weight. Additionally, transformation predicted values true scale reinforces correlation above 0.8 threshold.