A Hybrid Machine Learning Framework for Soil Classification in Smart Agriculture

Agriculture plays a central role in India’s economy, engaging
around 60% of the population and contributing significantly to the
national GDP [1]. However, the sector struggles with declining soil
quality, inefficient resource management, and a lack of timely, data
driven insights. Traditional soil classification methods are labor
intensive and imprecise, which limits effective agricultural planning. To
address these challenges, this study proposes a novel soil type
classification and prediction framework that integrates Internet of
Things (IoT) technology with advanced Machine Learning (ML)
algorithms to support smart farming practices.
The system collects real-time soil and environmental data using a suite
of IoT sensors—including pH, moisture, electrical conductivity,
temperature (DS18B20), and humidity (DHT11)—interfaced with an
ESP32 microcontroller. The gathered data is wirelessly transmitted to a
cloud platform, preprocessed through feature scaling and encoding, and
analyzed using multiple ML models.