Crop Recommendation: A Technology-Driven Approach to Enhancing Agricultural Productivity
Abstract
Agriculture is the lifeblood of farmers, and higher yield production that is in accordance with food security and market saturation ensuring high profitability is the achievement. The recommendation of crops in this context is a significant path to achieve this achievement. So, our concentration is on recommending the best-suited crop for a certain region, and hence, we proposed Stack Generalization(SG), a Strategic Ensemble ML based framework for recommending farmers the most appropriate crop for regional conditions. In this study, we used five basic models, namely DT, SVC, RF, KNN, and GB, and compared their performance with the proposed framework, which is SG (DT + SVC + RF + KNN + GB), based on Overall performance and Matthews Correlation Coefficient (MCC). For the analysis, we considered the essential soil nutrients NPK (nitrogen, phosphorus, and potassium) and the pH level and environmental traits THR (ie, temperature (in °C), rainfall (in mm) and humidity (in percent)), which are vital for soil fertility and crop development, and split the data as 80-20 percent. Our framework is undoubtedly capable of helping farmers in informed agricultural decision making, as the performance of the proposed schema in the means of overall accuracy 99. 77% with precision 99. 77%, recall 99. 83%, F1 score 99. 79% and MCC 99. 76% serves as evidence.