Files
Abstract
Species, in biology, are determined by a classification of related organisms that share common characteristics and are capable of interbreeding. People are interested in how species can be delimited, or whether their common ancestor can be found based on genetic sequences. In general, there are several methods used in species delimitation, such as the Automated Barcode Gap Discovery method, the General Mixed Yule Coalescent method, and the Poisson Tree Process method, etc. However, these methods have several disadvantages, including time consuming, hard to solve the big dataset problem, etc. In our design, we explore using supervised machine learning methods (Catboost, XGboost, Classification Tree, Support Vector Machine, K-nearest Neighbors) and an unsupervised machine learning method (K-means Clustering) in species delimitation. Five species trees are determined to be our treatments. The results show that supervised machine learning models have the highest accuracy compared to the unsupervised machine learning model.