A Framework for Data-Driven Agent-Based Modelling of Agricultural Land Use
Agent-based models (ABMs) are particularly suited for simulating the behaviour of agricultural agents in response to land use (LU) policy. However, there is no evidence of their widespread use by policymakers. Here, we carry out a review of LU ABMs to understand how farmers’ decision-making has been...
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2023-03-01
|
Series: | Land |
Subjects: | |
Online Access: | https://www.mdpi.com/2073-445X/12/4/756 |
_version_ | 1797604776219770880 |
---|---|
author | Giacomo Ravaioli Tiago Domingos Ricardo F. M. Teixeira |
author_facet | Giacomo Ravaioli Tiago Domingos Ricardo F. M. Teixeira |
author_sort | Giacomo Ravaioli |
collection | DOAJ |
description | Agent-based models (ABMs) are particularly suited for simulating the behaviour of agricultural agents in response to land use (LU) policy. However, there is no evidence of their widespread use by policymakers. Here, we carry out a review of LU ABMs to understand how farmers’ decision-making has been modelled. We found that LU ABMs mainly rely on pre-defined behavioural rules at the individual farmers’ level. They prioritise explanatory over predictive purposes, thus limiting the use of ABM for policy assessment. We explore the use of machine learning (ML) as a data-driven alternative for modelling decisions. Integration of ML with ABMs has never been properly applied to LU modelling, despite the increased availability of remote sensing products and agricultural micro-data. Therefore, we also propose a framework to develop data-driven ABMs for agricultural LU. This framework avoids pre-defined theoretical or heuristic rules and instead resorts to ML algorithms to learn agents’ behavioural rules from data. ML models are not directly interpretable, but their analysis can provide novel insights regarding the response of farmers to policy changes. The integration of ML models can also improve the validation of individual behaviours, which increases the ability of ABMs to predict policy outcomes at the micro-level. |
first_indexed | 2024-03-11T04:51:33Z |
format | Article |
id | doaj.art-f749d7fb426e47f581e150b58d4c784e |
institution | Directory Open Access Journal |
issn | 2073-445X |
language | English |
last_indexed | 2024-03-11T04:51:33Z |
publishDate | 2023-03-01 |
publisher | MDPI AG |
record_format | Article |
series | Land |
spelling | doaj.art-f749d7fb426e47f581e150b58d4c784e2023-11-17T20:01:44ZengMDPI AGLand2073-445X2023-03-0112475610.3390/land12040756A Framework for Data-Driven Agent-Based Modelling of Agricultural Land UseGiacomo Ravaioli0Tiago Domingos1Ricardo F. M. Teixeira2MARETEC-Marine, Environment and Technology Centre, LARSyS, Instituto Superior Técnico, Universidade de Lisboa, Av. Rovisco Pais, 1, 1049-001 Lisbon, PortugalMARETEC-Marine, Environment and Technology Centre, LARSyS, Instituto Superior Técnico, Universidade de Lisboa, Av. Rovisco Pais, 1, 1049-001 Lisbon, PortugalMARETEC-Marine, Environment and Technology Centre, LARSyS, Instituto Superior Técnico, Universidade de Lisboa, Av. Rovisco Pais, 1, 1049-001 Lisbon, PortugalAgent-based models (ABMs) are particularly suited for simulating the behaviour of agricultural agents in response to land use (LU) policy. However, there is no evidence of their widespread use by policymakers. Here, we carry out a review of LU ABMs to understand how farmers’ decision-making has been modelled. We found that LU ABMs mainly rely on pre-defined behavioural rules at the individual farmers’ level. They prioritise explanatory over predictive purposes, thus limiting the use of ABM for policy assessment. We explore the use of machine learning (ML) as a data-driven alternative for modelling decisions. Integration of ML with ABMs has never been properly applied to LU modelling, despite the increased availability of remote sensing products and agricultural micro-data. Therefore, we also propose a framework to develop data-driven ABMs for agricultural LU. This framework avoids pre-defined theoretical or heuristic rules and instead resorts to ML algorithms to learn agents’ behavioural rules from data. ML models are not directly interpretable, but their analysis can provide novel insights regarding the response of farmers to policy changes. The integration of ML models can also improve the validation of individual behaviours, which increases the ability of ABMs to predict policy outcomes at the micro-level.https://www.mdpi.com/2073-445X/12/4/756policy assessmentmachine learningbehavioural modellingdecision-making |
spellingShingle | Giacomo Ravaioli Tiago Domingos Ricardo F. M. Teixeira A Framework for Data-Driven Agent-Based Modelling of Agricultural Land Use Land policy assessment machine learning behavioural modelling decision-making |
title | A Framework for Data-Driven Agent-Based Modelling of Agricultural Land Use |
title_full | A Framework for Data-Driven Agent-Based Modelling of Agricultural Land Use |
title_fullStr | A Framework for Data-Driven Agent-Based Modelling of Agricultural Land Use |
title_full_unstemmed | A Framework for Data-Driven Agent-Based Modelling of Agricultural Land Use |
title_short | A Framework for Data-Driven Agent-Based Modelling of Agricultural Land Use |
title_sort | framework for data driven agent based modelling of agricultural land use |
topic | policy assessment machine learning behavioural modelling decision-making |
url | https://www.mdpi.com/2073-445X/12/4/756 |
work_keys_str_mv | AT giacomoravaioli aframeworkfordatadrivenagentbasedmodellingofagriculturallanduse AT tiagodomingos aframeworkfordatadrivenagentbasedmodellingofagriculturallanduse AT ricardofmteixeira aframeworkfordatadrivenagentbasedmodellingofagriculturallanduse AT giacomoravaioli frameworkfordatadrivenagentbasedmodellingofagriculturallanduse AT tiagodomingos frameworkfordatadrivenagentbasedmodellingofagriculturallanduse AT ricardofmteixeira frameworkfordatadrivenagentbasedmodellingofagriculturallanduse |