Predicting climate smart agriculture (CSA) practices using machine learning: A prime exploratory survey
The paper aim and novelty is the development of technology-based tools able of providing realistic insights on farmers’ future adaptation decisions by developing an ML algorithm to predict Climate-Smart Agriculture (CSA) practices and highlight modeling challenges to account for. And proposing a the...
| Autores principales: | , |
|---|---|
| Formato: | Journal Article |
| Lenguaje: | Inglés |
| Publicado: |
Elsevier
2024
|
| Materias: | |
| Acceso en línea: | https://hdl.handle.net/10568/141876 |
| _version_ | 1855538735065595904 |
|---|---|
| author | Noma, Freddy Babu, Suresh Chandra |
| author_browse | Babu, Suresh Chandra Noma, Freddy |
| author_facet | Noma, Freddy Babu, Suresh Chandra |
| author_sort | Noma, Freddy |
| collection | Repository of Agricultural Research Outputs (CGSpace) |
| description | The paper aim and novelty is the development of technology-based tools able of providing realistic insights on farmers’ future adaptation decisions by developing an ML algorithm to predict Climate-Smart Agriculture (CSA) practices and highlight modeling challenges to account for. And proposing a theoretical approach that grounds the selection of data (i.e. input and response variables) with well stablished theories on adaptation decision making process; with the aim of demonstrating ways of improving data science and ML publication quality in the field of agricultural economics. Data used are farmers’ socio-economic characteristics, farms’ features, agro-ecology’s features, climate indicators (temperature, rain, etc.), etc. In this paper, the optimized Gradient Boosting ML was trained and tested using households’ level data from Rakai district in Central Region of Uganda. The modeling approach was framed in climate adaptation analytical frameworks. Data extracted allows generating CSA clusters giving two response variables, used separately to train two different algorithms. The developed CSA predictive algorithm demonstrates that adaptation practices can be predicted using households’ level parameters. And both models are revealed to have fair performance metrics, with algorithm reaching up to 60% of accuracy. To further improve accuracy scores, deep-learning algorithms are suggested in future research. The developed CSA prediction algorithm could be used at both households and value chain levels, to select appropriate adaptation strategies, to plan adaptation, to estimate adaptation costs and develop investment’ plans. |
| format | Journal Article |
| id | CGSpace141876 |
| institution | CGIAR Consortium |
| language | Inglés |
| publishDate | 2024 |
| publishDateRange | 2024 |
| publishDateSort | 2024 |
| publisher | Elsevier |
| publisherStr | Elsevier |
| record_format | dspace |
| spelling | CGSpace1418762025-10-26T12:56:54Z Predicting climate smart agriculture (CSA) practices using machine learning: A prime exploratory survey Noma, Freddy Babu, Suresh Chandra climate-smart agriculture data farmers machine learning soil management water management machine learning The paper aim and novelty is the development of technology-based tools able of providing realistic insights on farmers’ future adaptation decisions by developing an ML algorithm to predict Climate-Smart Agriculture (CSA) practices and highlight modeling challenges to account for. And proposing a theoretical approach that grounds the selection of data (i.e. input and response variables) with well stablished theories on adaptation decision making process; with the aim of demonstrating ways of improving data science and ML publication quality in the field of agricultural economics. Data used are farmers’ socio-economic characteristics, farms’ features, agro-ecology’s features, climate indicators (temperature, rain, etc.), etc. In this paper, the optimized Gradient Boosting ML was trained and tested using households’ level data from Rakai district in Central Region of Uganda. The modeling approach was framed in climate adaptation analytical frameworks. Data extracted allows generating CSA clusters giving two response variables, used separately to train two different algorithms. The developed CSA predictive algorithm demonstrates that adaptation practices can be predicted using households’ level parameters. And both models are revealed to have fair performance metrics, with algorithm reaching up to 60% of accuracy. To further improve accuracy scores, deep-learning algorithms are suggested in future research. The developed CSA prediction algorithm could be used at both households and value chain levels, to select appropriate adaptation strategies, to plan adaptation, to estimate adaptation costs and develop investment’ plans. 2024-04 2024-05-16T19:48:53Z 2024-05-16T19:48:53Z Journal Article https://hdl.handle.net/10568/141876 en Open Access Elsevier Noma, Freddy; and Babu, Suresh Chandra. 2024. Predicting climate smart agriculture (CSA) practices using machine learning: A prime exploratory survey. Climate Services 34(April 2024): 100484. https://doi.org/10.1016/j.cliser.2024.100484 |
| spellingShingle | climate-smart agriculture data farmers machine learning soil management water management machine learning Noma, Freddy Babu, Suresh Chandra Predicting climate smart agriculture (CSA) practices using machine learning: A prime exploratory survey |
| title | Predicting climate smart agriculture (CSA) practices using machine learning: A prime exploratory survey |
| title_full | Predicting climate smart agriculture (CSA) practices using machine learning: A prime exploratory survey |
| title_fullStr | Predicting climate smart agriculture (CSA) practices using machine learning: A prime exploratory survey |
| title_full_unstemmed | Predicting climate smart agriculture (CSA) practices using machine learning: A prime exploratory survey |
| title_short | Predicting climate smart agriculture (CSA) practices using machine learning: A prime exploratory survey |
| title_sort | predicting climate smart agriculture csa practices using machine learning a prime exploratory survey |
| topic | climate-smart agriculture data farmers machine learning soil management water management machine learning |
| url | https://hdl.handle.net/10568/141876 |
| work_keys_str_mv | AT nomafreddy predictingclimatesmartagriculturecsapracticesusingmachinelearningaprimeexploratorysurvey AT babusureshchandra predictingclimatesmartagriculturecsapracticesusingmachinelearningaprimeexploratorysurvey |