Machine Learning has a strong potential in automated diagnosis of various diseases. With the recent upscale in various liver diseases, it is necessary to identify the liver disease at a preliminary stage. In this paper, we propose a new classifier by extending the XGBoost classifier with genetic algorithm. This paper compares various classification models and visualization techniques used to predict liver disease with feature selection. Outlier detection is used to find out the extreme deviating values and they are eliminated using isolation forest. The performance is measured in terms of accuracy, precision, recall f-measure and time complexity. The results of various classifiers are obtained by using proposed feature selection algorithm. From the experiments and comparative analysis, it increases classification accuracy and also leads to reduction in classification time and hence aids in the prediction of the disease more efficiently. © 2020 IEEE.