Record Details

NextGen agroadvisory expanding in scope and extent: The effort to cover more crops and bundle with lime advisory

CGSpace

View Archive Info
 
 
Field Value
 
Title NextGen agroadvisory expanding in scope and extent: The effort to cover more crops and bundle with lime advisory
 
Creator Tilaye, Asmalu
Abera, Wuletawu
Liben, Feyera
Ali, Ashenafi
Assefa, Feben
Tibebe, Degefie
Ebrahim, Mohammed
Mesfin, Tewodros
Erkossa, Teklu
Chernet, Meklit
Tamene, Lulseged
 
Subject agronomic practices
advisory services
modelling
decision support systems
bundling
random forest, agroadvisory, nextgen, site-specific
 
Description Accurate crop yield prediction is crucial for optimizing agricultural practices and ensuring food security. The NextGen advisory is a fertilizer recommendation system that can be customized for different farming systems, crop types, and locations. The system uses machine learning models to predict site-specific fertilizer rates using filed trial and covariates (soil, topography, and climate) data. In this study, we investigated the effectiveness of four machine learning models – random forest, support vector machine (SVM), k-nearest neighbor (KNN), and classification and regression tree (CART) – in predicting crop yields for barley, maize, and teff in Ethiopia. We employed repeated cross-validation with 10 folds and 3 repeats for each model to evaluate their performance. The models were assessed using three metrics: mean absolute error (MAE), root mean square error (RMSE), and R-square (R2). Our evaluation demonstrated that the random forest model outperformed the other models for all crops based with an R-square with training 074, 0.74, 0.71 and testing 0.74, 0.76, and 0.72 for barely, maize and teff respectively. This suggests that the random forest algorithm effectively captured the complex relationships between input features and crop yield. We are currently collecting additional data on crop response to fertilizer for barley, maize, and teff, as well as other crops. This additional data will be incorporated into the model to further enhance its predictive capabilities. Additionally, the model's performance will be validated in the 2023/2024 season in collaboration with government and private sector actors. Digital Green, Lersha, and the Ministry of Agriculture (MoA) have expressed interest in piloting the advisory service on smaller sites while the validation process is ongoing. The system is also being used to develop site-specific lime recommendations for acidic soils. The lime advisory tool can help to improve fertilizer use efficiency, reduce fertilizer costs, enhance soil health, and reduce the environmental impact of agriculture. These benefits can lead to sustainable crop production and profit from investment in inorganic and organic fertilizers.
 
Date 2023-11-26
2023-12-04T12:46:39Z
2023-12-04T12:46:39Z
 
Type Report
 
Identifier Tilaye, A.; Abera, W.; Liben, F.; Ali, A.; Assefa, F.; Tibebe, D.; Ebrahim, M.; Mesfin, T.; Erkossa, T.; Chernet, M.; Tamene, L. (2023) NextGen agroadvisory expanding in scope and extent: The effort to cover more crops and bundle with lime advisory. 13 p.
https://hdl.handle.net/10568/134945
 
Language en
 
Rights CC-BY-4.0
Open Access
 
Format 13 p.
application/pdf