The paper aims to summarize key research works in symbolic regression and perform a comparative study to understand the strengths and limitations of each method. Finally, we highlight the challenges in the current methods and future research directions in the application of machine learning in knowledge discovery.