KOMPARASI ALGORITMA KLASIFIKASI DATA MINING UNTUK MEMPREDIKSI TINGKAT KEMATIAN DINI KANKER DENGAN DATASET EARLY DEATH CANCER

Abstract

Cancer is something big in the world. Cancer is a malignant disease that is difficult to cure if the spread is too wide. However, detecting cancer cells as early as possible can reduce the risk of death. This study aims to predict the level of early detection of disasters in European countries using 5 classification algorithms, namely: Desecion Tree, Naïve Bayes, k-Nearset Neighbor, Random Forest and Neural Network of which algorithm is the best for this study. Tests carried out with several stages of research include: dataset (data contains), initial data processing, proposed method, credit method using 10 times cross validation, test results and t-test different tests. The alpha value is 0.05. if the probability is> 0.05 then H0 is accepted. If the probability is <0.05 then Ho is rejected. The results of the research that obtained performance with an accuracy value of 98.35% were the Neural Network algorithm. Whereas, the results of the research using the algirtic t-test with the best models are: Random Forest algorithm and Neural Network, the relatively good Naïve Bayes algorithm, the Desecion Tree algorithm is quite good and the poor algorithm is the K-Nearset Neighbor (K-NN) algorithm.