Predictive machine learning has emerged as a transformative technique in structural engineering nowadays, facilitating advanced analysis of intricate behaviour under axial compressive loads and buckling scenarios, more specifically of cold-formed steel structural elements. The current review of past studies examines modern machine learning applications for predicting these structures, emphasizing their capacity to model non-linear relationships, handle extensive and varied datasets and improve prediction accuracy relative to conventional analytical and numerical techniques. Supervised learning methods including support vector machines, random forests and artificial neural networks have proven to show considerable effectiveness in precisely forecasting the capacity of structural elements. Researchers can employ machine learning approaches to optimize the design process, reduce computational costs, and enhance security features in structural systems. The present research highlights the integration of Machine Learning (ML) and Finite Element Analysis (FEA) to generate datasets for training and validation that ensure dependable model performance. A comprehensive analysis of the importance of feature engineering, encompassing material properties, geometric parameters, and loading conditions, is provided, as these components are crucial for improving the model's generalization and interpretability. Despite promising advancements, challenges like as overfitting, data scarcity, and computer resource demands remain significant barriers to the broader implementation of these strategies. The increasing interest in machine learning models incorporating domain knowledge to enhance predictive accuracy and reduce data requirements is emphasised.