The use of decision tree analysis for improving age estimation standards from the acetabulum

Accurate, reliable and easy-to-use statistical methods in multifactorial age estimation from the skeleton remains a much-debated issue. In this paper, we explore the use of decision trees in adult age estimation. For this purpose, a dataset from 100 acetabula of South Africans, previously used for age estimation using transition analysis, were used to build a basic decision tree. A test sample of 25 individuals were then employed to assess the newly developed decision tree. Using the decision tree, 20 of the 25 individuals were classified into the correct age group (young, middle or older adults), with the remaining 5 falling within the adjacent age group. The decision tree provided a more accurate outcome as compared to the previous study using transition analysis. Although much research is still needed, this analysis suggests that decision trees may be usable in adult age estimation and may handle the non-linear relationship between chronological and biological age somewhat better than other traditional statistical methods.