The primary objective of the present study was to assess the accuracy of various forecasting methods for predicting the unconfined compressive strength (UCS) of fine-grained soils in the Fewa Lake region. This assessment was conducted through a comprehensive analysis involving theoretical, statistical, and computational approaches. Additionally, the study aimed to evaluate the effectiveness and performance of newly developed machine learning (ML) models in this context. To gauge the precision of the developed models, several statistical metrics were employed. These included the correlation coefficient (r), coefficient of determination (R2), mean absolute error (MAE), mean absolute percentage error (MAPE), weighted mean absolute percentage error (WMAPE), and Nash–Sutcliffe Efficiency (NSE). By utilizing these metrics, the study quantified the accuracy of the predictive models. One of the pivotal findings of this research underscores the notable importance of machine learning techniques in accurately forecasting the UCS of fine-grained soil. However, it was also revealed that the performance of these machine learning models can be influenced by factors such as multicollinearity and limited database size. Despite these challenges, the study significantly contributes to the advancement of soil assessment practices, particularly for samples of fine-grained soil in the Fewa Lake region.