LIVER DISEASE PREDICTION USING SVM AND NAÏVE BAYES ALGORITHMS

Abstract

When it comes to automated illness detection and prediction, data mining is a crucial 
component. The process uses data mining algorithms and methods to examine health records. 
One of the leading causes of death in a number of nations is liver disease, which has been on 
the rise in recent years. Predicting liver disease using classification models built from datasets 
of liver patients is the goal of this thesis. This thesis improved the prediction accuracy of 
Indian liver patients in three stages by constructing feature models and comparing them. The 
first step is applying the min max normalization technique to the original datasets of liver 
patients obtained from the UCI repository. During the second step of liver dataset prediction, 
a subset of the normalized liver patient dataset is produced using PSO feature selection. This 
subset contains just the most important features. In the third step, the data set is subjected to 
categorization methods. In the last stage, we'll figure out how accurate it is by calculating the 
root mean square and root mean error values. After using PSO feature selection, the J48 
method is thought to be the superior algorithm in terms of performance. The assessment is 
concluded by looking at the accuracy values.