Cone penetration test-based assessment of liquefaction potential using machine and hybrid learning approaches

Multiscale and Multidisciplinary Modeling Experiments and Design, Год журнала: 2024, Номер 7(4), С. 3841 - 3864

Опубликована: Апрель 26, 2024

Язык: Английский

---

The effectiveness of data pre-processing methods on the performance of machine learning techniques using RF, SVR, Cubist and SGB: a study on undrained shear strength prediction

Stochastic Environmental Research and Risk Assessment, Год журнала: 2024, Номер 38(8), С. 3273 - 3290

Опубликована: Июнь 13, 2024

Abstract In the field of data engineering in machine learning (ML), a crucial component is process scaling, normalization, and standardization. This involves transforming to make it more compatible with modeling techniques. particular, this transformation essential ensure suitability for subsequent analysis. Despite application many conventional relatively new approaches ML, there remains conspicuous lack research, particularly geotechnical discipline. study, ML-based prediction models (i.e., RF, SVR, Cubist, SGB) were developed estimate undrained shear strength (UDSS) cohesive soil from perspective wide range data-scaling methods. Therefore, work presents novel ML framework based on Cubist regression method predict UDSS soil. A dataset including six different features one target variable used building models. The performance was examined considering impact pre-processing issue. For that purpose, scaling methods, namely Range, Z-Score, Log Transformation, Box-Cox, Yeo-Johnson, generate results then systematically compared using sampling ratios understand how model varies as various scaling/transformation methods algorithms combined. It observed or had considerable limited effects depending algorithm type ratio. Compared SGB models, provided higher metrics after applying steps. Box-Cox transformed yielded best among other an R 2 0.87 90% training set. Also, generally when transformed-based Log, Yeo-Johnson) than scaled-based Range Z-Score) show has potential prediction, have impacts predictive capacity evaluated

Язык: Английский

---

Improving Aggregate Abrasion Resistance Prediction via Micro-Deval Test Using Ensemble Machine Learning Techniques

Engineering Journal, Год журнала: 2024, Номер 28(3), С. 15 - 24

Опубликована: Март 1, 2024

Aggregate is the most extracted material from world's mines and widely used in civil construction projects.The Micro-Deval abrasion test (MD) one of important tests that provides characteristics crushed aggregates show their resistance against mechanical abrasive factors such as repeated impact loading.The various on properties has led researchers to seek correlations, often focusing limited data samples, leading reduced accuracy.This study employs machine learning (ML) methods predict MD values, considering diverse aggregate properties.Various ensemble ML were applied, revealing exceptional performance stacking model, which achieved an R 2 score 0.95 predicting resistance.The feature importance analysis highlights influence Magnesium Sulfate Soundness (MSS), Water Absorption (ABS), Los Angeles Abrasion (LAA) suggesting use multiple could yield a more dependable assessment durability.

Язык: Английский

---

Application of state-of-the-art machine learning algorithms for slope stability prediction by handling outliers of the dataset

Earth Science Informatics, Год журнала: 2023, Номер 16(3), С. 2497 - 2509

Опубликована: Июль 24, 2023

Язык: Английский

---

Prediction of collapsibility of loess site based on artificial intelligence: comparison of different algorithms

Environmental Earth Sciences, Год журнала: 2024, Номер 83(3)

Опубликована: Янв. 25, 2024

Язык: Английский

---