Crop Prediction Using Machine Learning and IoT: A Comparative Analysis of Algorithms

Agriculture is a cornerstone of Bangladesh's econ-omy, yet it faces significant challenges due to weather fluctuations, including changes in temperature, humidity, and precipitation. These environmental variables critically affect farming activities, demanding constant adaptation. To mitigate these challenges, we propose an loT-based model designed to enhance agricultural data collection and provide valuable insights to farmers. Our model integrates various sensors, such as the ESP8266-0l mi-crocontroller, NPK sensor, DHT22 humidity sensor, DSl8B20 temperature sensor, pH sensor, and water level sensor, all housed in a protective enclosure and powered by a DC source. This system collects real-time data on critical environmental parameters, forming the foundation of predictive models. We employ multiple machine learning algorithms, including Logistic Regression, Naive Bayes, Support Vector Machine, K-Nearest Neighbours, Decision Tree, Random Forest, Bagging, AdaBoost, Gradient Boosting, and Extra Trees. Using a dataset containing parameters like nitrogen, phosphorus, potassium, temperature, humidity, pH, and rainfall, we split the data into training and testing sets. Models are evaluated based on accuracy, precision, recall, Fl score, and undergo l0-fold cross-validation to assess performance comprehensively. The integration of loT-derived data with machine learning models helps identify the most suitable crops based on soil and climate conditions. By offering real-time, actionable insights, our system enables farmers to make better crop selection decisions. This approach aims to enhance agricultural productivity in Bangladesh by optimizing weather-dependent farming strategies and ensuring more informed agri-cultural practices.