In observational astronomy, one of the primary challenges is to accurately identify stars that belong to a particular cluster or those that are part of field stars based solely on photometric data. This identification process is crucial as it serves as the initial step to isolate the target objects and precisely analyze the properties of a star cluster. Recently, machine learning algorithms and astrometric data have been utilized to tackle this problem. This study aims to compare the effectiveness of various unsupervised machine learning methods in the membership association of stars in open and globular clusters. A dataset consisting of three open and three globular star clusters from GAIA DR3 have been used to test the viability of these methods on different types of clusters. The study analyzed the sensitivity, precision, and false-discovery rate of methods such as GMM, DBSCAN, and pyUPMASK. The results showed that pyUPMASK had the highest precision, albeit with slightly higher computational time, while GMM and DBSCAN performed similarly. This study highlights the importance of selecting the appropriate machine-learning method to estimate the membership probability of a star belonging to a cluster based on cluster type, size, and composition.