A MACHINE LEARNING FRAMEWORK FOR PRECISION CROP YIELD PREDICTION AND OPTIMIZATION

Information Technology

The Machine Learning-Based Crop Yield Prediction Model is an innovative application designed to assist farmers, agricultural experts, and policymakers by accurately predicting crop yields based on environmental and agricultural factors. This system utilizes machine learning models to analyze data such as soil properties, weather conditions, and farming practices, providing actionable insights for better decision-making. The model is trained using historical crop data in CSV format, which includes features such as temperature, rainfall, humidity, soil pH, and nutrient levels. It will give fertilizers based on crops and its features like temperature, humidity Preprocessing techniques are applied to clean and normalize the data, ensuring high-quality inputs for training. Algorithms such as Linear Regression, Random Forest, or Gradient Boosting are used to predict crop yield, offering reliable and accurate results. The existing system for analyzing crop productivity in Tamil Nadu considers key factors like rainfall, groundwater, cultivation area, and soil type to optimize the yield of crops such as rice, maize, ragi, sugarcane, and tapioca. Data from the past ten years is collected from various sources, including government websites, and transformed into a suitable format for analysis. The system uses K-Means clustering and classification algorithms such as fuzzy logic, KNN, and Modified KNN, with MKNN providing the best prediction results. Additionally, IoT devices like soil sensors, DHT11 sensors for humidity and temperature, and Arduino Uno with Atmega processors collect real-time atmospheric data. The Naïve Bayes algorithm achieves 97% accuracy, further enhanced using a boosting algorithm for improved performance. Regression techniques like ENet, Kernel Ridge, and Lasso are employed, with Stacking Regression for superior yield prediction.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.