Share
Export Citation
A New Model Application of AIoT-Based Crop Recommendation and Monitoring System
Subiyanto
International Journal on Engineering Applications
Q2Abstract
Modern agriculture demands accurate approaches integrating Artificial Intelligence of Things (AIoT) technology to optimize crop growth under dynamic environmental conditions. Although current recommendation systems exist, they lack the accuracy to support effectively farmers in making data-driven decisions. This work presents an accurate recommendation system by using AIoT technology. The framework begins with the collection of essential environmental data including nitrogen (N), phosphorus (P), potassium (K), pH, temperature, and humidity using soil-embedded sensors, with an ESP32 microcontroller managing data transmission as a gateway. Then this real-time data is processed on a cloud platform and analyzed through a stacking ensemble machine learning model, combining algorithms such as K-Nearest Neighbors, Decision Trees, Random Forest Classifier, Gradient Boosting Classifier, Support Vector Machine, and Artificial Neural Networks, with XGBoost selected as the meta-learner to optimize prediction accuracy. The proposed model demonstrates high performance, achieving an accuracy of 0.999, along with precision, recall, F1-score, and ROC-AUC metrics nearing perfection at 0.999, underscoring its robustness in precision agriculture applications. Furthermore, the real-time data acquisition and monitoring enabled by the AIoT framework ensure timely recommendations, advancing agricultural productivity and sustainability.