Showing 1–2 of 2 results for author: Podpečan, V

Explainable machine learning for predicting shellfish toxicity in the Adriatic Sea using long-term monitoring data of HABs

Authors: Martin Marzidovšek, Janja Francé, Vid Podpečan, Stanka Vadnjal, Jožica Dolenc, Patricija Mozetič

Abstract: In this study, explainable machine learning techniques are applied to predict the toxicity of mussels in the Gulf of Trieste (Adriatic Sea) caused by harmful algal blooms. By analysing a newly created 28-year dataset containing records of toxic phytoplankton in mussel farming areas and toxin concentrations in mussels (Mytilus galloprovincialis), we train and evaluate the performance of ML models t… ▽ More In this study, explainable machine learning techniques are applied to predict the toxicity of mussels in the Gulf of Trieste (Adriatic Sea) caused by harmful algal blooms. By analysing a newly created 28-year dataset containing records of toxic phytoplankton in mussel farming areas and toxin concentrations in mussels (Mytilus galloprovincialis), we train and evaluate the performance of ML models to accurately predict diarrhetic shellfish poisoning (DSP) events. The random forest model provided the best prediction of positive toxicity results based on the F1 score. Explainability methods such as permutation importance and SHAP identified key species (Dinophysis fortii and D. caudata) and environmental factors (salinity, river discharge and precipitation) as the best predictors of DSP outbreaks. These findings are important for improving early warning systems and supporting sustainable aquaculture practices. △ Less

Submitted 9 May, 2024; v1 submitted 7 May, 2024; originally announced May 2024.

BEFANA: A Tool for Biodiversity-Ecosystem Functioning Assessment by Network Analysis

Authors: Martin Marzidovšek, Vid Podpečan, Erminia Conti, Marko Debeljak, Christian Mulder

Abstract: BEFANA is a free and open-source software tool for ecological network analysis and visualisation. It is adapted to ecologists' needs and allows them to study the topology and dynamics of ecological networks as well as apply selected machine learning algorithms. BEFANA is implemented in Python, and structured as an ordered collection of interactive computational notebooks. It relies on widely used… ▽ More BEFANA is a free and open-source software tool for ecological network analysis and visualisation. It is adapted to ecologists' needs and allows them to study the topology and dynamics of ecological networks as well as apply selected machine learning algorithms. BEFANA is implemented in Python, and structured as an ordered collection of interactive computational notebooks. It relies on widely used open-source libraries, and aims to achieve simplicity, interactivity, and extensibility. BEFANA provides methods and implementations for data loading and preprocessing, network analysis and interactive visualisation, modelling with experimental data, and predictive modelling with machine learning. We showcase BEFANA through a concrete example of a detrital soil food web of agricultural grasslands, and demonstrate all of its main components and functionalities. △ Less

Submitted 28 March, 2022; v1 submitted 21 March, 2022; originally announced March 2022.