Crop Recommendation System

Abstract

Ensuring food security gets tricky with climate change and limited resources. That means we need smart, data-based tech in agriculture. The research investigates a crop recommendation system which depends on machine learning technology. The system operates independently from any particular dataset requirements. The system recommends suitable crops through analysis of soil characteristics together with environmental factors including temperature and humidity and rainfall patterns. Regular systems stick to set datasets and number-based soil details. This one handle category inputs for soil, like sandy, loamy, or clayey. It can run on any structured data in CSV or Excel format. The setup includes a strong preprocessing step. That covers normalizing data, encoding it, and standardizing features. All this helps it adapt to different datasets. We tried out several algorithms for this. Those include Random Forest, Decision Tree, K-Nearest Neighbors, and Gaussian Naive Bayes. We checked how they performed. The results show Naive Bayes hit 99.32 percent accuracy. Random Forest came in at 99.09 percent. Both did really well. The system tackles issues with varying data pretty effectively. It provides a way to scale up that is smart and easy for users. Farmers get practical suggestions on crops from it. In the end, this pushes forward precision agriculture and ways to farm sustainably.