<![CDATA[Classification of skull bones by sex is part of human biological profile identification in forensic anthropology that aims to determine whether the skeleton belongs to a male or female. The most popular method for determining sex from bones is DNA analysis. However, under some conditions such as burnt, damaged, or very dry skeletal remains, DNA analysis cannot provide accurate results. So forensic anthropology is developing by utilizing the help of machine learning technology. This research shows the performance of Support Vector Machine in classifying skull bones based on gender. The skull parameter data used is data collected by Dr. William Howells from craniometric measurements consisting of male and female data with a total of 2524 data and 82 features, namely bizygomatic breadth, glabello-occipital lenght and others. In building the skull bone classification model, the Support Vector Machine kernels used are linear, RBF, and polynomial. Based on the test results, the best accuracy was obtained in each kernel function, namely the linear kernel obtained the best accuracy of 88.14% with C = 2. For the RBF kernel, the best accuracy was 91.30% at C = 2, γ = 'auto'. For the polynomial kernel, the best accuracy was 88.14% at C = 1 and 2, γ = 1 and 2, d = 1. The evaluation results show that the Support Vector Machine model with the RBF kernel has proven to be the optimal choice in skull bone classification compared to other kernels, based on accuracy, precision, recall, and CrossValidation measurements reaching values above 90%. These results indicate that the skull bone classification model based on gender using Support Vector Machine is recommended in forensic anthropology.]]>
Copyrights © 2024
