5–15% of women worldwide was found diagnosed with Polycystic Ovary Syndrome, creating an adverse effect on their reproductive health through hormonal imbalances, ovarian dysfunction, and metabolic irregularities. Hence, early identification and appropriate treatment are crucial to improve the patient’s health. This study explores ML algorithms such as Support Vector Machine, Decision Tree, Logistic Regression, and K-Nearest Neighbor, to classify the presence and absence of PCOS using dataset of 539 clinical entries, later compared with a subset of 136 entries to assess the performance differences. This research aims to identify significant factors contributing to PCOS and predict infertility risks. Results showed varying accuracies across models, with ordinal logistic regression achieving notable success in predicting infertility risk. The findings underscore the relationship between PCOS severity and infertility risk, highlighting the need for tailored intervention strategies. This work advances PCOS research by emphasizing the potential of machine learning in early detection and effective risk stratification, paving the way for better clinical management worldwide.