A data-augmented machine learning framework for estimating undrained mechanical properties of a cement-improved clay

This repository contains the dataset and training codes used in a data-augmented machine learning study for estimating undrained mechanical properties of a cement-improved clay. The dataset consists of 540 samples, including 30 experimental observations obtained from laboratory testing and 510 synthetically generated samples produced to enhance model training. The dataset includes cement content, effective consolidation pressure, specimen density, undrained shear strength, and undrained stiffness modulus. The accompanying code comprises a Jupyter Notebook implementing artificial neural network and random forest models for predicting undrained shear strength and undrained stiffness modulus from relevant input parameters. The repository is organised into separate data and code directories. The notebook is provided in a cleared state and can be executed sequentially using relative paths. Detailed information on dataset structure, model configurations, and execution instructions is provided in the included README.md file. This dataset and code are intended to support transparency, reproducibility, and reuse in geotechnical engineering and machine learning research.