Fertilizer Recommendation for Agriculture using Machine Learning

Ineffective fertilizer use frequently reduces agricultural output by degrading soil, increasing expenses and producing less-than-ideal crop yields. Conventional fertilizer application techniques focus on broad guidelines and ignore particular soil and environmental circumstances, leading to deficiencies in nutrients and resource waste. This study suggests a machine learning-based fertilizer recommendation system that offers precise, data-driven recommendations based on specific agricultural conditions in order to address this problem. Utilizing exploratory data analysis (EDA) and processing techniques the system analyzes important variables like temperature, humidity, moisture, crop kind soil type and vital macronutrients like potassium, phosphorus and nitrogen to guarantee high-quality input. After testing several classifiers such as Random Forest (which overfits at 100%) and KNN (93.4%), a Decision Tree Classifier with an accuracy of 99.78% is used as the final model. Farmers can enter soil and environmental characteristics in real time to receive accurate fertilizer recommendations instantaneously thanks to the system's integration with a Streamlit-based user interface. This technology supports efficient and healthy farming by optimizing the use of fertilizers, increasing crop production, decreasing nutrient waste and encouraging sustainable agricultural practices.