CROP YIELD PREDICTION SYSTEM USING MACHINE LEARNING

Agricultural productivity plays a pivotal role in the economy and food security of India, where over 50% of the population is involved in farming. However, due to changing climatic conditions, inconsistent rainfall patterns, and lack of precise planning tools, farmers often struggle with crop selection and expected yield. This project introduces a Smart Indian Crop Yield Prediction System that leverages Machine Learning (ML) algorithms and real-time weather data to forecast the yield of a selected crop based on soil nutrients, area, and weather conditions. Using historical agricultural data, the system is trained to learn relationships between parameters such as Nitrogen (N), Phosphorus (P), Potassium (K) content in the soil, area under cultivation, and seasonal and crop type factors. The system incorporates real-time temperature and humidity data using a weather API, enhancing the accuracy of predictions. Users (farmers or agriculture officers) can input their location, crop type, and soil properties through a web-based interface. The system then processes this data using pre-trained ML models such as Linear Regression, Random Forest, or Decision Tree, depending on the user’s selection. Key outputs include: predicted yield per hectare, visualization of evaluation metrics (MAE, MSE, RMSE, R²), and actual vs predicted yield plots. Users can also download a detailed prediction report. The proposed system serves as a decision-support tool aimed at improving agricultural productivity, reducing risk, and supporting data-driven decision-making in rural areas. It also provides scalability by integrating more parameters like rainfall, soil pH, or fertilizer dosage in future iterations.