PCOS DISEASE CLASSIFICATION USING FEATURE SELECTION RFECV AND EDA WITH KNN ALGORITHM METHOD

Abstract

Polycystic ovary syndrome is an endocrine disorder of the ovaries that causes hormonal disturbances in women of reproductive age, where androgen secretion in the ovaries of women with Polycystic Ovary Syndrome (PCOS) is excessive compared to normal women. This usually occur in women with obesity which is characterized by irregular menstrual cycles, chronic anovulation, hyperandrogenism, and even infertility. Efforts are used to treat this disease in the form of hormone therapy, laparoscopic ovarian drilling, and in-vitro fertilization. However, these three therapies are focused on symptomatic therapy and are less effective in treating PCOS-related infertility. Detecting PCOS disease early is very necessary so that prevention and treatment can be carried out immediately. Therefore, a classification is carried out to detect PCOS disease by being able to analyze data that has a high degree of accuracy. The method used for the classification of PCOS disease is using the K Nearest Neighbor (KNN), method which previously carried out the feature selection process, namely the Exploratory Data Analysis (EDA), method which is used for the data analysis process by means of an analysis approach to data to find out the most accurate method and using the Recursive Feature Elimination and Cross-Validation (RFECV) selection method which ranks the features based on their level of importance to the prediction process. Further, the data classification process uses the K-Nearest Neighbors (KNN) algorithm. The results of the Exploratory Data Analysis (EDA) feature selection process produce 10 data attributes that are used and are continued by the Recursive Feature Elimination and Cross-Validation (RFECV) process by producing the 7 most important attributes used and finally the K-Nearest Neighbors (KNN) method has a high level high accuracy by producing an accuracy value of 93%, precision 82%, recall 100%, and F1 score 90%.