KLASIFIKASI MULTI-LABEL MENGGUNAKAN METODE MULTI-LABEL K-NEAREST NEIGHBOR (ML-KNN) PADA PENYAKIT KANKER SERVIKS

Abstract

Based on GLOBOCAN 2020 statistical data, cervical cancer ranks 8th most cancers suffered by women worldwide with 604,127 cases and 341,831 deaths. Meanwhile, in Indonesia, cervical cancer sufferers are in 2nd place with 36,633 cases with a death rate of 21,003 people. Multi-Label K-Nearest Neighbor (ML-KNN) is an adaptive algorithm that can be used to solve multi-label classification cases. This research uses a dataset obtained from the UCI Machine Learning website. The dataset will be pre-processed by deleting missing values, checking for duplicate data, checking data types, and resample data in the form of oversampling on the Biopsy label due to unbalanced class 1 and 0 data. Furthermore, the data is divided into training data and test data with a ratio of 80:20. In the training data, look for its proximity to the predetermined k value, namely K=1, K=3, K=5, K=7, and K=9. The best performance evaluation results were obtained when the value of K = 5 which obtained a hamming loss value of 3.59%, accuracy of 93%, precision weighted of 93%, recall weighted of 96%, and f1-score weighted of 94%.