Performance Evaluation of Heterogeneous Ensemble Methods for Polycystic Ovarian Syndrome Detection

Polycystic Ovarian Syndrome (PCOS), a hormonal condition, is one of the most common causes of female infertility where the ovaries frequently form tiny cysts which produces multiple symptoms from irregular menstruation cycle and associated fertility issues. PCOS affects approximately 10 million people worldwide. Many studies involve detecting PCOS were developed involving machine learning and PCOS classification. In order to further expand the findings of the field, the study evaluates the performance of Heterogenous Ensemble methods, mainly Stacking and Voting methods, for PCOS detection. By using Data Augmentation and Feature Selection, the Stack Ensemble PCOS Classifier was the highest performing heterogenous ensemble method for PCOS detection with accuracy, precision, recall and f1 score of 92.72%, 93.15%, 92.38%, and 92.69% respectively.