DEVELOPMENT OF A MARKET PRICE PREDICTION SYSTEM FOR FARMERS USING DECISION TREE

ABSTRACT The market price prediction system serves family and smallholder farmers by enabling them to make informed decisions by providing real-time predictions and updates on agricultural commodity prices to maximize profit while reducing risk. Numerous farmers remain without access to forecasting systems that assist their crop planning and sales approach. This project delivers a market price prediction system, particularly designed for farmers' assistance. The platform offers a user-friendly and easy-to-navigate system that relies on a modern development architecture that unites Next.js for front-end development with Tailwind CSS for styling, Jupyter Notebook for machine learning training, Uvicorn for local server creation, Python for backend server, FastAPI for API route creation, Joblib for saving and loading the machine learning model, scikit-learn for implementing the machine learning algorithm, and Prisma, joined with MongoDB, to run backend data processing. The Decision Tree algorithm powers the predictive system because it receives training from historical agricultural commodity price data spanning the period from 2017 to 2023. Using this platform, farmers can sign up, log in, submit product information, which enables them to view and analyze predicted pricing patterns, and here they also receive instant notifications on price trends, together with a discussion area for community cooperation. Since the system uses machine learning and is user-friendly, it gives farmers helpful insights, increases the transparency of markets, and supports efficient agricultural plans as well as good use of resources. A solution that implements easy-to-use technology practices alongside explainable machine learning functionality adds value to the agricultural sector while protecting agricultural financial sustainability.