Classification of Heavy Metal-Indicated Soil Types Using Decision Tree Algorithm and IoT-Based pH-Moisture Sensors

This study implements the Decision Tree algorithm for the classification of three soil types: Paddy Field, Well Excavation, and Lapindo Mud. The classification is performed based on two key parameters that influence soil characteristics: pH and Humidity. The dataset consists of 120 samples, partitioned using a ratio of 75% training data (90 samples) and 25% testing data (30 samples). The model testing results on the unseen data demonstrated very high performance, achieving an overall Accuracy of 93.33%. The analysis of performance metrics indicates that the model possesses strong discriminatory power across all classes. The Lapindo Mud class achieved perfect Recall (1.00), primarily driven by its unique alkaline pH (pH > 7.0) which acts as the root node separator. Furthermore, the Well Excavation class demonstrated perfect Precision (1.00), indicating absolute reliability in its predictions. The F1-Score for all classes exceeded 0.90. These results confirm that the Decision Tree Model is an efficient and effective method for identifying and classifying soil types based on pH and Humidity parameters.