Use of Machine Learning Techniques in Soil Classification

In the design of reliable structures, soil classification process is first step, which involves costly and time-consuming work including laboratory tests. Machine learning (ML), has wide use in many scientific fields, can be utilized for facilitating classification. This study aims to provide a concrete example ML The dataset comprises 805 samples based on drillings new Gayrettepe–Istanbul Airport metro line construction. both missing data class imbalance. preprocessing stage, first, imputation techniques were applied deal with data. Two different tested, finally, imputed KNN imputer. Later, balance was achieved synthetic minority oversampling technique (SMOTE). After preprocessing, series algorithms tested 10-fold cross-validation. Unlike studies conducted previous research, gradient-boosting methods such as XGBoost, LightGBM, CatBoost high accuracy rates up +90% observed, significant improvement prediction (when compared research) achieved.