Use of the Naive Bayes Function and the Models of Artificial Neural Networks to Classify Some Cancer Tumors Al-Sabbah Shrook A.S.1,*, Mohammad Sada Faydh1, Eanad Maryam Mahdi1 1Statistics Department, Administration and Economics College, Kerbala University, Iraq *Corresponding author: Shrook A.S. Al-Sabbah, E-mail: shorouq.a@uokerbala.edu.iqmean
Online published on 30 April, 2019. Abstract This study is concerned with the methods of classification and separation of observations and the use of two methods: of classification Naive Bayes, using the function of Kernell, and models of artificial neural networks using a function to show the step or sample in order to classify three types of cancer tumors (bone, lung, and breast). For a simple random sample of 150 patients, Naive Bayes was the correct classification of data for 67%, while in the method of neural network models, it was 86%. The model of artificial neural networks was found to be the best in classifying views from the Naive Bayes classification. The correct classification of bone cancer is the highest, followed by breast cancer and lung cancer. The relative importance of the variable age was the highest, then sex, and then the period of survival of the patient, and then the profession of the patient, and then the state of exit of the patient. Top Keywords Naive Bayes, Neural Networks, Kernell Function. Top |