An innovative kernel-based recursive time-series learning algorithm with applications to improvements of beehive management practices

Abstract

In this paper, we propose an innovative kernel-based learning algorithm to sequentially estimate subset vector autoregressive models (including full-order models). To demonstrate the effectiveness of the proposed recursive algorithm, we apply this algorithm to test the direct causal relationships between the population of honeybee foragers and foraging types gathering nectar, pollen or water. We have found that under certain conditions, nectar foraging may be improved by the changes in the proportions of other foraging bees, such as pollen foragers. This suggests that we may be able to predict the optimal conditions at any time to maximise the honey yield of colonies.