Analisis Resiko Kanker Serviks Menggunakan PCA-ANFIS Berdasarkan Historical Medical Record

Abstract

Cervical cancer is the second highest disease in the category of women that accounts for the most deaths based on data from WHO. Based on data from the Global Cancer Observatory (Globocan), cervical cancer has a total of 36,633 cases with a death rate of 21,003 cases due to cervical cancer and is relatively high. People with this disease are often difficult to distinguish between healthy and not. So the purpose of this study is to discuss the diagnosis of cervical cancer using the Adaptive Neuro Fuzzy Inference System (ANFIS) and Principal Component Analysis (PCA) reduction to find the best accuracy in the diagnosis of cervical cancer. ANFIS is a system used for modeling based on fuzzy Sugeno, which considers the simplicity of computation. PCA is a method applied to express information expressed in data and specified in an alternative form. The amount of data used is 72 data with 10 features. Then the data is normalized and feature reduction is performed using PCA. After doing feature reduction by PCA obtained 4 influential features. Furthermore, analysis using ANFIS was carried out from the data that PCA extraction was carried out and which was not carried out. Then the accuracy test is carried out using the confusion matrix. The best result based on diagnostic accuracy is ANFIS using 91.67% PCA with the model obtained from the 3rd k-fold and the membership function type is trapmf, while the accuracy without PCA is smaller than using PCA, which is 86.67%.