Machine learning methods typically focus on the correlations within data while neglecting the causal relationships that reveal underlying mechanisms. This limitation may restrict the reliability and interpretability of models in decision support and intervention strategies. For this reason, causal discovery methods have gained widespread attention. They can infer causal structures and directions between variables from observational data, thereby providing decision-makers with an interpretable and intervenable analytical framework. This review introduces commonly used causal discovery methods based on observational data. Combined with specific case studies, it demonstrates and practices these methods using the R language, aiming to provide readers with practical references for understanding and applying causal discovery methods.