Species identification using pairwise LDA.

Comparison of LDA (linear discriminant analysis) classification accuracy of each pairwise model. The list of “tooth related” and “non-tooth related” features is given in Table 2. Two-tailed -tests show that the mean accuracy of each of the three feature sets is significantly higher than the previous set, or significantly higher than a random classifier in the first case ( in all cases).本研究对比了各成对模型的线性判别分析（LDA）分类准确率。表2列出了“牙齿相关特征”与“非牙齿相关特征”的明细清单。经双侧t检验证实，三类特征集各自的平均准确率均显著高于前一类特征集；针对首个特征集，其平均准确率显著高于随机分类器，且所有场景均符合该规律。