Prediction of Soil Compression Index using SVM and kNN

Abstract Soil is used both as a building material and as the surface on which construction is carried out hence this demonstrates how important soil is as a material. Therefore, it is vital to evaluate the soil’s characteristics, such as its strength and settlement, before any construction is built on the soil. The most crucial factor that must be established in order to compute soil settlement is compression index, which can be obtained through a laboratory oedometer consolidation test. Numerous empirical correlations were created because the oedometer test is challenging and takes time for determining this parameter. With the development of technology, it has become much simpler to forecast the compression index parameter using a variety of other easy-to-find soil parameters. In this study, the prediction of the compression index has been attempted utilizing machine learning methods such as support vector machines and k-nearest neighbors. In order to forecast the output parameter, compression index, machine learning models use soil index qualities including liquid limit, plasticity index, natural moisture content, and void ratio as the input parameter. There are 359 total data used for analysis from data acquired from various studies. Typically, machine learning models divide the data into training and testing datasets in order to train the model and forecast its performance. As a result, different ratios of data splitting are also utilized when developing the machine learning model. Using measures like mean square error, mean absolute error and correlation coefficient, the model’s performance is assessed. Additionally, this paper also examines the impact of the model-creating parameters.