Implementing and Analyzing RNN for Predicting Nutrient Requirements in Agricultural Products

The project's goal was to utilize advanced machine learning methods, specifically deep learning, to offer practical guidance on identifying key nutrients essential for agricultural plant growth. These nutrients include minerals like nitrogen, potassium, and phosphorus, affected by environmental conditions including temperature, soil acidity, and rainfall. The data used in this study, sourced from Kaggle, helped in giving actionable recommendations for choosing appropriate crops and determining nutrient needs for specific locations. The study applied a Recurrent Neural Network (RNN) to predict essential nutrients for plant growth using Kaggle-sourced data on environmental factors and nutrients. The RNN effectively captured temporal patterns, achieving an accuracy of about 64 % in predicting nitrogen, phosphorus, and potassium requirements. Training showed a consistent increase in accuracy, peaking at 200–300 epochs, confirming the model's effectiveness. The experimental results demonstrated a model accuracy of approximately 63% on the test set, with a training loss of 0.25, An average squared error of 0.14 and an absolute mean error of 2.14%. These findings highlight the RNN's potential to improve nutrient management and crop selection, supporting more efficient and sustainable agricultural practices.