EUVIMEDEUVIMED entdecken
Crossref

FROM PREDICTION TO OPTIMIZATION: MACHINE LEARNING AND GENETIC ALGORITHMS FRAMEWORK FOR AGRO-ECOSYSTEM MANAGEMENT

Alketa Hyso, Dezdemona Gjylapi

International Journal of Ecosystems and Ecology Science (IJEES) · 19.06.2026 · 16 (3) · 107-116

DOI: 10.31407/ijees16.314 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal of Ecosystems and Ecology Science (IJEES), online erschienen am 19.06.2026. Verfasst von Alketa Hyso und 1 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This study presents an integrated framework for agro-ecosystem decision support that combines machine learningbased crop-yield prediction with Genetic Algorithm (GA)-based optimization of controllable farm-management variables.Preliminary experiments on the original simulated dataset showed weak learnable structure and predictive performance close to a naive baseline.To address this limitation, a crop-aware synthetic dataset was constructed for methodological validation, expanding the data from 500 to 1,030 observations across five crop types.The dataset introduced a more coherent agronomic response structure linking soil, climate, vegetation, management, and disease factors to crop yield.The analysis included several tree-based ensemble models, such as Random Forest and Extra Trees, as well as boosting methods including Gradient Boosting, XGBoost, and CatBoost.The best-performing pipeline, EnhancedCatBoostLog, achieved a test MAE of 235.73,RMSE of 295.56, and R2 of 0.9358.The optimization stage focused on soil moisture, soil pH, pesticide usage, irrigation type, and fertilizer type, while maintaining sunlight exposure as a fixed environmental input.A five-fold GA evaluation using repeated model retraining and fold-level test samples yielded an average improvement of 263.90 kg/ha, corresponding to 12.38%, with positive gains in 97.5% of cases.These results show the methodological potential of integrating predictive modeling and resource-aware optimization for agro-ecosystem management under controlled synthetic conditions.

Zitieren

APA

Hyso, A., & Gjylapi, D. (2026). FROM PREDICTION TO OPTIMIZATION: MACHINE LEARNING AND GENETIC ALGORITHMS FRAMEWORK FOR AGRO-ECOSYSTEM MANAGEMENT. International Journal of Ecosystems and Ecology Science (IJEES), 16(3), 107-116. https://doi.org/10.31407/ijees16.314

Vancouver

Hyso A, Gjylapi D. FROM PREDICTION TO OPTIMIZATION: MACHINE LEARNING AND GENETIC ALGORITHMS FRAMEWORK FOR AGRO-ECOSYSTEM MANAGEMENT. International Journal of Ecosystems and Ecology Science (IJEES). 2026;16(3):107-116. doi:10.31407/ijees16.314

BibTeX
@article{Hyso2026,
  author = {Hyso, Alketa and Gjylapi, Dezdemona},
  title = {FROM PREDICTION TO OPTIMIZATION: MACHINE LEARNING AND GENETIC ALGORITHMS FRAMEWORK FOR AGRO-ECOSYSTEM MANAGEMENT},
  journal = {International Journal of Ecosystems and Ecology Science (IJEES)},
  year = {2026},
  volume = {16},
  number = {3},
  pages = {107--116},
  doi = {10.31407/ijees16.314}
}
Kostenlos starten →