Data-driven Modelling of the Climatic Impact on the Microbial Food S afety of Raw Milk

The food sector is expected to be drastically affected by climate change. The main focus has been on the impact of climate change on food security and less on food safety. The focus of this paper is on dairy science and, in particular, the microbial food safety of raw cow milk. To assess the climate change effect, the characterization of the climatic impact is required. The goal of this paper is to develop models that capture the relationship between several climate variables, e.g., temperature, etc., and the microbial food safety of raw milk, in terms of total bacterial counts levels measured in CFU/mL. The data used originate from 123 Maltese dairy farms over 6 years. The developed models evaluate the incorporation of temporal aspects on both linear approaches, i.e., partial least squares and an altered version of partial least squares regression with auto-regressive elements, and non-linear approaches, i.e., artificial neural networks and long short-term memory recurrent networks. Results show that linear methods overall provide better accuracy. The potential of the applied methods to develop models that are suitable to be utilised as impact models is demonstrated. Such impact models are useful to conduct climate change impact assessments to evaluate future microbial food safety risks due to climate change for dairy products.