Florida scientists use AI to predict when Burmese pythons are most likely to be caught

University of Florida researchers have employed machine learning to create a Weighted Removal Index (WRI) that can predict when Burmese python removal surveys are most likely to capture individuals with the greatest potential to influence population growth. By incorporating demographic information about individual snakes, the WRI provides a more accurate assessment of the removals' ecological value compared to traditional methods that simply count captured snakes.

The study, published in Ecological Applications, highlights that large reproductive females and warmer, more humid conditions with waning lunar phases are associated with higher survey values. The neural network model performed the best at predicting survey success, achieving a balanced accuracy of 0.703 and lowest prediction error for survey WRI.

The findings suggest that managers should direct their resources towards snake removal periods that are more likely to have a meaningful impact on the invasive population.