Speciation Prediction based on Spatial Distribution and Spatiotemporal Information from an Individual-Based Ecosystem Simulation

DOI: 10.5176/2251-2179_ATAI22

Authors: Morteza Mashayekhi and Robin Gras

Abstract:

In this paper, by using machine learning techniques, we are going to investigate the ability of species’ spatial and spatiotemporal distribution information in an individual-based ecosystem simulation (Ecosim) for speciation prediction. Because of the imbalanced nature of our dataset we use smote algorithm to make a relatively balanced dataset to avoid dismissing the minor class samples. Experimental results show very good results for the test set generated from the same run as the learning set. It also shows good results on test sets generated from different runs of Ecosim. We also observe superior results when we use, for the learning set, a run with more species compare to a run with less species. Finally we noticed that spatial information is very effective in speciation prediction and spatiotemporal information can improve it.

Keywords: speciation; spacial distribution; spatiotemporal information; speciation prediction; smote