Enhancing Agricultural Yield: A Unified Stacking Ensemble Method for Crop Recommendations using Soil Properties and Weather Attributes

Abstract

Agriculture plays a remarkably pivotal role in propelling India's economic growth. In the Indian context, the agricultural sector generates more employment opportunities compared to any other domain. It accounts for half of the entire workforce in our nation. However, owing to their limited understanding and lack of awareness about diverse soil constituents and environmental elements, farmers often make incorrect choices regarding the crops they cultivate. This misjudgment significantly hampers agricultural productivity. The creation of a systematic approach that offers scientific guidance to farmers in predicting optimal crops for cultivation based on various factors influencing overall production becomes crucial to surmount this challenge. Accurate crop forecasting holds paramount importance for enhancing agricultural techniques, mitigating risks, ensuring food security, and promoting sustainable farming practices. Throughout the agricultural value chain, such forecasting enables well-informed decision-making and efficient allocation of resources by furnishing insightful insights into crop yields and production. In this endeavor, a dataset incorporating soil parameters like Nitrogen (N), Phosphorous (P), Potassium (K), and soil pH, as well as weather parameters encompassing temperature, rainfall, and humidity for 22 different crops, was employed. As an enhancement derived synthetic data from original data set to bring in novelty in producing larger dataset. Advanced ensemble stacking machine learning techniques were applied to recommend crops with remarkable accuracy and efficiency. Hence, this approach holds the potential to grant farmers greater flexibility and versatility in their agricultural pursuits.