PCOS, or Polycystic Ovary Syndrome, is a frequent hormonal condition affecting women during their reproductive years. This disorder results in irregular and infrequent menstrual cycles, which can cause infertility and other related health issues. PCOS can present with various symptoms such as irregular periods, acne, obesity, and excessive hair growth, among others. However, some symptoms may not be visibly apparent and can only be detected through testing. This paper focuses on different techniques and algorithms and compares them to suggest which model is best suited to accurately classifying whether a woman has PCOS or not. The algorithms Random Forest Classifier (RFC), Support Vector Machine (SVM), XGBoost, and Ensemble Learning are applied on the basis of dimensionality reduction and Principal Component Analysis (PCA) to datasets available on Kaggle. This dataset consists of 43 attributes for 541 women, among whom 177 have PCOS. Based on the accuracy, it was found that the XGBoost Classifier performed better than the other models and ended up giving an overall accuracy of 89.63% before applying PCA and that RFC and SVC performed similarly and better than the other models and ended up giving an overall accuracy of 91.11% after applying PCA and retaining 99% of its variance.
Paper
Full text
A Comparative Study on Different Machine Learning Algorithms to Detect PCOS
Semantic Scholar · Medicine · 2023
Abstract
PCOS, or Polycystic Ovary Syndrome, is a frequent hormonal condition affecting women during their reproductive years. This disorder results in irregular and infrequent menstrual cycles, which can cause infertility and other related health issues. PCOS can present with various symptoms such as irregular periods, acne, obesity, and excessive hair growth, among others. However, some symptoms may not be visibly apparent and can only be detected through testing. This paper focuses on different techniques and algorithms and compares them to suggest which model is best suited to accurately classifying whether a woman has PCOS or not. The algorithms Random Forest Classifier (RFC), Support Vector Machine (SVM), XGBoost, and Ensemble Learning are applied on the basis of dimensionality reduction and Principal Component Analysis (PCA) to datasets available on Kaggle. This dataset consists of 43 attributes for 541 women, among whom 177 have PCOS. Based on the accuracy, it was found that the XGBoost Classifier performed better than the other models and ended up giving an overall accuracy of 89.63% before applying PCA and that RFC and SVC performed similarly and better than the other models and ended up giving an overall accuracy of 91.11% after applying PCA and retaining 99% of its variance.