AI-BASED DECISION SUPPORT SYSTEMS FOR PRECISION AGRICULTURE MANAGEMENT

Abstract

Precision agriculture is becoming more towards the data-driven intelligent to deal with variability of soil, crop health, climate and resource availability, but the current practices of decision-making are inadequately disjointed, reactive, and overly reliant on the heuristics of farmers. This paper considers the issue of ineffective and slow agricultural decisions offering an AI-based Decision Support System (DSS) to manage precision agriculture. The first goal is to create a common system that combines the multiple sources of agricultural information and some sophisticated machine learning models to assist in making timely, precise, and understandable farm-level choices. In the proposed DSS, the models that are used to analyze the data of soil moisture, weather conditions, crop health indices, and previous yield data include Random Forest (RF), Support Vector Machine (SVM), Gradient Boosting (XGBoost), and Long Short-Term Memory (LSTM). The data collection is carried out using the IoT sensors and remote sensors, feature engineering, model training and validation as well as decision rule creation to be used in irrigation, fertilization and disease control. The experimental findings prove that AI-based DSS provides a high level of prediction accuracy and operational efficiency. XGBoost has demonstrated the best overall accuracy of 96.8 %and is superior to RF (94.1%), SVM (92.6%), and LSTM (95.2%). The system took up less water by 18.7 %, less fertilizer was used by 14.3 %, and there was a 21.5 % reduction in error in yield prediction as compared to traditional rule based methods. The results attest to the fact that AI-based DSS allows taking a proactive, data-driven approach to agricultural management, improving productivity, sustainability, and resource optimization in precision farming models.