Abstract
The aim of the study was to optimise the parameters of a neural network model for forecasting plant pest population dynamics in agrocenoses with variable environmental characteristics. The methodology included weekly field data collection over five months at 18 sites in northern Kazakhstan, taking into account six pest species, environmental parameters, crop characteristics, and pest control schemes. Data cleaning, linear interpolation, normalisation, and one-hot encoding were applied. A multilayer perceptron (MLP) architecture with two hidden layers (32 and 16 neurons) and Rectified Linear Unit (ReLU) activation, trained using various configurations, was used as the baseline. The results showed that the highest values of the determination coefficient (R²>0.82) were achieved when modelling Helicoverpa armigera and Frankliniella occidentalis, while the poorest performance was observed for Loxostege sticticalis (R²≈0.64). In plots with tomatoes and sunflowers, forecast accuracy was higher, which was associated with stable irrigation conditions and protective measures. The average Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) values were 4.36 and 3.52, respectively. The optimal configuration was identified as a two-layer model (64 and 32 neurons) with ReLU activation and the Adaptive Moment Estimation (ADAM) optimiser, yielding MAE=2.9 and R²=0.83. Deeper models led to overfitting, whereas simplified ones resulted in insufficient accuracy. Statistical tests confirmed the significance of differences between configurations (p<0.001). Cross-validation and sliding window analysis demonstrated the model’s high robustness to spatiotemporal variability, with minimal accuracy degradation when transferred across regions and seasons. The results demonstrate the high practical value of the developed approach for integration into pest monitoring and forecasting systems in agricultural practice.
Abstract from the paper’s OpenAlex record; open copy licensed cc-by. Read at the source ↗
The text the model read, with the source record’s inline markup (italics, sub- and superscripts) rendered and its character entities decoded; quote offsets refer to the raw text.
Claims 14
Claim · model-written summary
Helicoverpa armigera forecasts attained the highest R², above 0.82.
- Taxon
- Helicoverpa armigera
_gusano_cogollero.png?width=60)
Helicoverpa armigerae0 - Outcome
- determination coefficiente0
- Direction
- higheste0
- Value
- R²>0.82e0
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
_gusano_cogollero.png?width=500)
Verbatim
e0“the highest values of the determination coefficient (R²>0.82) were achieved when modelling Helicoverpa armigera and Frankliniella occidentalis”
Claim · model-written summary
Frankliniella occidentalis forecasts attained the highest R², above 0.82.
- Taxon
- Frankliniella occidentalis
Frankliniella occidentalise0 - Outcome
- determination coefficiente0
- Direction
- higheste0
- Value
- R²>0.82e0
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
e0“the highest values of the determination coefficient (R²>0.82) were achieved when modelling Helicoverpa armigera and Frankliniella occidentalis”
Claim · model-written summary
Loxostege sticticalis forecasts performed poorest, at R² approximately 0.64.
- Taxon
- Loxostege sticticalis
Loxostege sticticalise0 - Outcome
- performancee0
- Direction
- pooreste0
- Value
- R²≈0.64e0
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
e0“the poorest performance was observed for Loxostege sticticalis (R²≈0.64).”
Claim · model-written summary
Forecast accuracy was higher in tomato plots, associated with stable irrigation and protection.
- Taxon
- tomatoes
Solanum lycopersicume0 - Treatment
- stable irrigation conditions and protective measurese0
- Outcome
- forecast accuracye0
- Direction
- highere0
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
e0“In plots with tomatoes and sunflowers, forecast accuracy was higher, which was associated with stable irrigation conditions and protective measures.”
Claim · model-written summary
Forecast accuracy was higher in sunflower plots, associated with stable irrigation and protection.
- Taxon
- sunflowers
Helianthuse0 - Treatment
- stable irrigation conditions and protective measurese0
- Outcome
- forecast accuracye0
- Direction
- highere0
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
e0“In plots with tomatoes and sunflowers, forecast accuracy was higher, which was associated with stable irrigation conditions and protective measures.”
Claim · model-written summary
Average forecast RMSE was 4.36.
Not reportedTaxonDirection
Verbatim
e0“The average Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) values were 4.36 and 3.52, respectively.”
Claim · model-written summary
Average forecast MAE was 3.52.
Not reportedTaxonDirection
Verbatim
e0“The average Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) values were 4.36 and 3.52, respectively.”
Claim · model-written summary
The optimal two-layer 64/32-neuron ReLU ADAM model yielded MAE 2.9.
- Outcome
- MAEe0
- Value
- 2.9e0
- Qualifier
- two-layer model (64 and 32 neurons) with ReLU activation and the Adaptive Moment Estimation (ADAM) optimisere0
Not reportedTaxonDirection
Verbatim
e0“The optimal configuration was identified as a two-layer model (64 and 32 neurons) with ReLU activation and the Adaptive Moment Estimation (ADAM) optimiser, yielding MAE=2.9 and R²=0.83.”
Claim · model-written summary
The optimal model yielded R² 0.83.
Not reportedTaxonDirection
Verbatim
e0“The optimal configuration was identified as a two-layer model (64 and 32 neurons) with ReLU activation and the Adaptive Moment Estimation (ADAM) optimiser, yielding MAE=2.9 and R²=0.83.”
Claim · model-written summary
Deeper models overfit.
Not reportedTaxonComparator
Verbatim
e0“Deeper models led to overfitting, whereas simplified ones resulted in insufficient accuracy.”
Claim · model-written summary
Simplified models had insufficient accuracy.
- Outcome
- insufficient accuracye0
Not reportedTaxonComparator
Verbatim
e0“Deeper models led to overfitting, whereas simplified ones resulted in insufficient accuracy.”
Claim · model-written summary
Configurations differed significantly, p below 0.001.
Not reportedTaxonDirection
Verbatim
e0“Statistical tests confirmed the significance of differences between configurations (p<0.001).”
Claim · model-written summary
Validation demonstrated high robustness with minimal accuracy loss across regions and seasons.
- Outcome
- high robustness to spatiotemporal variabilitye0
- Qualifier
- minimal accuracy degradation when transferred across regions and seasonse0
Not reportedTaxonComparator
Verbatim
e0“Cross-validation and sliding window analysis demonstrated the model’s high robustness to spatiotemporal variability, with minimal accuracy degradation when transferred across regions and seasons.”
Claim · model-written summary
The approach has high practical value for agricultural pest monitoring and forecasting integration.
- Outcome
- integration into pest monitoring and forecasting systems in agricultural practicee0
Not reportedTaxon
Verbatim
e0“The results demonstrate the high practical value of the developed approach for integration into pest monitoring and forecasting systems in agricultural practice.”