Evaluation Of Machine Learning Models with Functional Selection of Climate-Based Harvest Recommendation Systems

The implementation of machine learning techniques to optimize crop selection based on essential soil and climate parameters. The soil attributes analyzed include pH values, nitrogen (S), phosphorus (P), and potassium (K) concentrations, while climate variables include temperature (°C), moisture (%) and precipitation (mm). By using the dataset for harvest recommendations from Kaggle, the study identifies key determinants of plant suitability and evaluates the effectiveness of several machine learning algorithms, such as accidental forests, gradient increase, vector machine (SVM) support, and KNearest Neighbor (KNN). The aim is to improve agricultural productivity through data-controlled recommendations tailored to specific environmental conditions. This study uses advanced feature selection methods to identify the primary variables that have an impact on harvest recommendations and provide valuable insight into the key factors needed for accurate predictions. Various models of the evaluation of machine learning are done using power metrics such as accuracy, recall, and F1 scores to determine the most effective approach. The findings indicate that all models compared to other algorithms exhibit strong prediction skills, with gradients and random forests consistently increasing higher accuracy and reliability. This result provides a valuable perspective for the continued development of precision agriculture and its practical implementation, which promotes a data-controlled approach to optimizing crop selection.