APPLICATION OF A HYBRID MACHINE LEARNING MODEL FOR SOIL TYPE CLASSIFICATION

This article presents a hybrid machine learning model designed for soil type classification based on the analysis of geophysical characteristics. The proposed model combines two algorithms – RandomForestClassifier and MLPClassifier – integrating the high accuracy of ensemble methods with the ability of neural networks to capture complex nonlinear dependencies between parameters. The input dataset included indicators such as electrical conductivity, density, P-wave propagation velocity, and burial depth. Prior to training, data preprocessing was performed, including outlier removal, standardization, and categorical feature encoding. The hybrid architecture allowed the integration of results from both models with different weights, optimizing classification accuracy. The effectiveness of the proposed approach was compared with alternative algorithms such as XGBoost and Keras using metrics including Accuracy, F1-score, Precision, and Recall. The hybrid model achieved an accuracy of 96.07%, outperforming individual algorithms. Visualization of confusion matrices provided insights into class distribution and model robustness. The results confirm that combining ensemble and neural methods ensures more stable and reliable predictions when working with geophysical data. The developed model can be effectively applied in geotechnical studies, construction, agriculture, and environmental monitoring, enhancing analytical efficiency and reducing the need for costly laboratory testing.

1. Aydın, A., Keskin, H., Öztürk, A. Use of Machine Learning Techniques in Soil Classification. Sustainability, 15 (3), 2374 (2023). https://doi.org/10.3390/su15032374.

2. Breiman, L. Random Forests. Machine Learning, 45(1), 5–32 (2001). https://doi.org/10.1023/A:1010933404324.

3. Zhang, W., Li, Y., Chen, X. Deep Learning-Based Soil Type Classification. Computers and Electronics in Agriculture, 169, 105205 (2020). https://doi.org/10.1016/j.compag.2019.105205.

4. Hengl, T., Heuvelink, G.B.M., Kempen, B. SoilGrids250m: Global Gridded Soil Information Based on Machine Learning/ PLoS ONE, 12 (2), e0169748 (2017). https://doi.org/10.1371/journal.pone.0169748.

5. Chlingaryan, A., Sukkarieh, S., Whelan, B. Machine Learning Approaches for Crop Yield Prediction and Soil Classification. Agricultural Systems, 167, 144–153 (2018). https://doi.org/10.1016/j.agsy.2018.09.012.

6. Gislason, P.O., Benediktsson, J.A., Sveinsson, J.R. Random Forests for Land Cover Classification. Remote Sensing of Environment, 110 (4), 435–449 (2006). https://doi.org/10.1016/j.rse.2007.03.012.

7. Pal, M. Random Forest Classifier for Remote Sensing Classification. International Journal of Remote Sensing, 26 (1), 217–222 (2005). https://doi.org/10.1080/01431160412331269698.

8. Schmidhuber, J. Deep Learning in Neural Networks: An Overview. Neural Networks, 61, 85–117 (2015). https://doi.org/10.1016/j.neunet.2014.09.003.

9. Vapnik, V. Statistical Learning Theory. New York: John Wiley & Sons, 1998.

10. Gandomi, A.H., Alavi A.H. A Review of Machine Learning Applications in Geotechnical and Geoenvironmental Engineering. Engineering Geology, 152, 63–81 (2013). https://doi.org/10.1016/j.enggeo.2013.05.016.