PCOS (polycystic ovarian syndrome) is a prevalent endocrine illness marked by irregular menstruation, hormonal imbalances, and metabolic problems, among women of reproductive age. In India, the prevalence of PCOS is estimated to be around 9–18%. For PCOS to be well managed and related consequences like obesity, heart disease, and infertility to be prevented, a timely and precise diagnosis is essential. Prevention of long-term problems may be achieved through early exposure and treatment. Machine learning techniques have shown promise in health care fields which also includes PCOS detection. Therefore, machine learning (ML) is applied to help with PCOS diagnosis. With the help of clinical features such as age, body mass ratio, and biochemical features, the model tries to diagnose PCOS in patients. In this work, the importance of feature selection and way to resampling the target features are emphasized to enhance the performance of the model. Various machine learning models are implemented and evaluated the effectiveness using performance metrics such as accuracy, precision, recall, and F1-score. In our paper, the Random Forest (RF) and Artificial Neural Network (ANN) perform better among other machine learning models implemented, with an accuracy of 91%. Further, using Bayesian model averaging, the high accuracy models are combined and the accuracy, 91.8% is achieved.