Detection of Hepatitis (A, B, C and E) Viruses Based on Random Forest, K-nearest and Naïve Bayes Classifier

Abstract

Recently, the techniques of data mining are broadly used to analyze biomedical data. These techniques have given efficient results in the prediction and classification of diseases severity and diagnosis of sicknesses. Hepatitis is a liver inflammation; it can affect people of all ages. Millions of people globally are thought to be affected by hepatitis. Accurate and early diagnosis of hepatitis can save many patients. Hepatitis is a key challenge for public health care services, due to limited clinical diagnosis of hepatitis disease in its early stages. This paper presents different techniques of data mining techniques used for diagnosing hepatitis and shows the results of different techniques in term of accuracy and training time. We used naive bayes, K-nearest and Random Forest classifier in WEKA software to calculate the result. Naïve bayes is the algorithm that used for solving text classification problems. K-nearest neighbor is a simple, supervised machine learning algorithm that is used for both regression and classification problem. Random Forest is one of the most used algorithms for its simplicity and the reason that it can be used for both regression and classification. In naïve bayes accuracy of our result can be found 93.2%. In Random Forest classifier the accuracy is 98.6% by using 10 fold cross-validation and in K-nearest neghbor it is become 95.8% by using 10 fold cross-validation.