{"id":{"repo_id":"helsinki","oai_identifier":"oai:helda.helsinki.fi:10138/229152"},"canonical_url":"https://search.dev.ndltd.org/etd/helsinki/oai:helda.helsinki.fi:10138/229152","repository":{"repo_id":"helsinki","name":"University of Helsinki","base_url":"https://helda.helsinki.fi/server/oai/request"},"display":{"title":"Modelling the Baltic Sea food web with a Dynamic Bayesian Network with hidden variables","abstract":"Protection of the functioning of ecosystems requires understanding of the ecological processes, which, however, are highly complex on multiple spatial and temporal scales. This complexity is a major challenge for modellers, particularly as ecological data are often scarce, and ecosystems are known to sometimes undergo relatively fast structural changes that have a major effect on the ecosystem dynamics. These changes may be driven by unobserved variables, i.e. ecosystem components that we do not have data on. This thesis fits a Dynamic Bayesian Network (DBN) model to one such ecosystem, the Baltic Sea. Three versions of a DBN of the Gotland Basin food web are fitted to data, to evaluate the role of different setup of hidden variables. The hidden variables were able to find similar patterns and links to observed variables in the time series regardless of their exact setup. The models predicted the last three years of the data rather poorly, which is probably due to a change in the time series exactly at the beginning of the predicted period. This indicates that while the hidden variables were able to pick up patterns in the data, there are still ecosystem changes that the model cannot predict.","abstract_html":"Protection of the functioning of ecosystems requires understanding of the ecological processes, which, however, are highly complex on multiple spatial and temporal scales. This complexity is a major challenge for modellers, particularly as ecological data are often scarce, and ecosystems are known to sometimes undergo relatively fast structural changes that have a major effect on the ecosystem dynamics. These changes may be driven by unobserved variables, i.e. ecosystem components that we do not have data on. This thesis fits a Dynamic Bayesian Network (DBN) model to one such ecosystem, the Baltic Sea. Three versions of a DBN of the Gotland Basin food web are fitted to data, to evaluate the role of different setup of hidden variables. The hidden variables were able to find similar patterns and links to observed variables in the time series regardless of their exact setup. The models predicted the last three years of the data rather poorly, which is probably due to a change in the time series exactly at the beginning of the predicted period. This indicates that while the hidden variables were able to pick up patterns in the data, there are still ecosystem changes that the model cannot predict.","abstract_has_math":false,"creators":["Uusitalo, Laura"],"institution":"Helsingfors universitet","degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":["Helsingin yliopisto, Matemaattis-luonnontieteellinen tiedekunta, Tietojenkäsittelytieteen laitos","University of Helsinki, Faculty of Science, Department of Computer Science","Helsingfors universitet, Matematisk-naturvetenskapliga fakulteten, Institutionen för datavetenskap"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2017,"date_issued":"2017","date_published":"2017","updated_at":"2026-07-27T19:56:17Z","subjects":[],"languages":["eng"],"rights":[],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["URN:NBN:fi-fe2017112251280"],"render_values":[{"text":"URN:NBN:fi-fe2017112251280","href":null,"code":true}]}]},"links":{"outbound_url":"http://hdl.handle.net/10138/229152","outbound_label":"Handle","outbound_source":"dc:identifier.uri"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Helsingin yliopisto, Matemaattis-luonnontieteellinen tiedekunta, Tietojenkäsittelytieteen laitos","University of Helsinki, Faculty of Science, Department of Computer Science","Helsingfors universitet, Matematisk-naturvetenskapliga fakulteten, Institutionen för datavetenskap"]},{"key":"dc:creator","label":"Author","values":["Uusitalo, Laura"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.issued","label":"Date","values":["2017"]},{"key":"dc:publisher","label":"Institution","values":["Helsingfors universitet","University of Helsinki","Helsingin yliopisto"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["eng"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["URN:NBN:fi-fe2017112251280","http://hdl.handle.net/10138/229152"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Protection of the functioning of ecosystems requires understanding of the ecological processes, which, however, are highly complex on multiple spatial and temporal scales. This complexity is a major challenge for modellers, particularly as ecological data are often scarce, and ecosystems are known to sometimes undergo relatively fast structural changes that have a major effect on the ecosystem dynamics. These changes may be driven by unobserved variables, i.e. ecosystem components that we do not have data on. This thesis fits a Dynamic Bayesian Network (DBN) model to one such ecosystem, the Baltic Sea. Three versions of a DBN of the Gotland Basin food web are fitted to data, to evaluate the role of different setup of hidden variables. The hidden variables were able to find similar patterns and links to observed variables in the time series regardless of their exact setup. The models predicted the last three years of the data rather poorly, which is probably due to a change in the time series exactly at the beginning of the predicted period. This indicates that while the hidden variables were able to pick up patterns in the data, there are still ecosystem changes that the model cannot predict."]},{"key":"dc:title","label":"Title","values":["Modelling the Baltic Sea food web with a Dynamic Bayesian Network with hidden variables"]}]}],"canonical_facts":{"dc:contributor":["Helsingin yliopisto, Matemaattis-luonnontieteellinen tiedekunta, Tietojenkäsittelytieteen laitos","University of Helsinki, Faculty of Science, Department of Computer Science","Helsingfors universitet, Matematisk-naturvetenskapliga fakulteten, Institutionen för datavetenskap"],"dc:creator":["Uusitalo, Laura"],"dc:date.issued":["2017"],"dc:description.abstract":["Protection of the functioning of ecosystems requires understanding of the ecological processes, which, however, are highly complex on multiple spatial and temporal scales. This complexity is a major challenge for modellers, particularly as ecological data are often scarce, and ecosystems are known to sometimes undergo relatively fast structural changes that have a major effect on the ecosystem dynamics. These changes may be driven by unobserved variables, i.e. ecosystem components that we do not have data on. This thesis fits a Dynamic Bayesian Network (DBN) model to one such ecosystem, the Baltic Sea. Three versions of a DBN of the Gotland Basin food web are fitted to data, to evaluate the role of different setup of hidden variables. The hidden variables were able to find similar patterns and links to observed variables in the time series regardless of their exact setup. The models predicted the last three years of the data rather poorly, which is probably due to a change in the time series exactly at the beginning of the predicted period. This indicates that while the hidden variables were able to pick up patterns in the data, there are still ecosystem changes that the model cannot predict."],"dc:identifier.uri":["URN:NBN:fi-fe2017112251280","http://hdl.handle.net/10138/229152"],"dc:language.iso":["eng"],"dc:publisher":["Helsingfors universitet","University of Helsinki","Helsingin yliopisto"],"dc:title":["Modelling the Baltic Sea food web with a Dynamic Bayesian Network with hidden variables"]},"updated_at":"2026-07-27T19:56:17Z"}