Abstract: Sunflower cultivation is a vital agricultural practice in India, supporting the livelihoods of more than 350,000 families. Since the emergence of sunflower rust disease in 1983, these families have faced substantial difficulties in maintaining crop yield and quality. This study seeks to create a robust sunflower yield prediction system using machine learning techniques, specifically focusing on Decision Tree, K-Nearest Neighbor (KNN), and Linear Regression algorithms. The system leverages a dataset that includes weather conditions, soil properties, and historical yield data from seven taluks in the Mysuru district. Data preprocessing steps, such as handling missing values and data normalization, ensure the dataset's integrity. The study evaluates the performance of Decision Tree, KNN, and Linear Regression in predicting sunflower yield, with an emphasis on accuracy, precision, and recall metrics. The findings reveal that Decision Tree and KNN, with their classification capabilities based on proximity to nearest neighbors, deliver more accurate predictions compared to Linear Regression, which models the linear relationships between variables. The resulting system serves as a practical tool for farmers, helping them make informed decisions regarding crop management and yield optimization. The study highlights the significant potential of integrating machine learning in agriculture, particularly in predicting crop yields and addressing challenges related to agricultural planning and resource management.
|
DOI:
10.17148/IJARCCE.2024.13852
