Estudo comparativo de modelos de machine learning para estimar a data de colheita no contexto da agricultura de precisão
Enregistré dans:
| Auteur principal: | |
|---|---|
| Date de publication: | 2025 |
| Format: | Bachelor thesis |
| Langue: | por |
| Source: | Repositório Institucional da UPF |
| Download full: | https://repositorio.upf.br/handle/123456789/9535 |
Résumé: | Accurately determining the harvest date is a strategic decision in agricultural management, directly impacting yield, grain quality, and crop sustainability. This study presents an exploratory investigation evaluating the feasibility of using machine learning algorithms to predict the harvest date of soybean and corn crops in the state of Rio Grande do Sul, Brazil. A dataset of one thousand records was built using real data and controlled simulations, including variables such as soil type, pH, nutrient levels (NPK), organic matter, temperature, humidity, precipitation, and solar radiation. Four regression models were implemented and compared: Linear Regression, Random Forest, MLP, and XGBoost. The analysis showed that Linear Regression achieved the best performance (MAE = 3.839172 days), outperforming even more complex models. The study also led to the development of a functional interface for data input and harvest prediction visualization. Results demonstrate that machine learning models can significantly support decision-making in agriculture, provided that the input data is reliable and representative. This research lays the groundwork for future studies and highlights promising paths for integrating artificial intelligence into agricultural planning. |