ANALISA METODE RANDOM FOREST TREE DAN K - NEAREST NEIGHBOR DALAM MENDETEKSI KANKER SERVIKS

STEELE, EDWARD SUWANDY SALIM, HARTATO BINDAN, ENDY PRANOTO (2020) ANALISA METODE RANDOM FOREST TREE DAN K - NEAREST NEIGHBOR DALAM MENDETEKSI KANKER SERVIKS , SKRIPSI, UNIVERSITAS PRIMA INDONESIA

ABSTRAK

Cervical cancer attacks the cell on patient cervix. Cervical cancer is ranked two as the cause of woman cancer in the whole world. From WHO (World Health Organization) data, 500.000 woman each year diagnosed with cervical cancer and 300.000 of those died. The mortality rate of cervical cancer is increasing throughout the year. The mortality rate of said disease can reach up to 60% and most of them are petitborgeois or lower cause they can’t afford to get early diagnosed. Hinselmann, Schiller, Citology, and Biopsy are four screening technics to diagnose whether there is cancer cell on patient cervix. In this research, the patient medical history dataset will be analyzed by using Random Forest Tree and KNN algorithm. Both of the algorithms will be compared to find the most accurate model to be implemented in order to figure out the patterns on cervical cancer patient and predicting the patient screening result whether it’s positive cervical cancer or negative. The result of this research is processed by using programming code Python for 214 test data from 854 total data. The accuracy at the end is shown 88.7% for Random Forest and 90.6% for KNN. Dataset used have four target classification which is a multilabel classification. KNN was proven to be more advance on its prediction of multilabel classification in detecting the patterns of cervical cancer patient’s case.

JURNAL