Liver Disease Prediction Using Support Vector Machine and Logistic Regression Model with Combination of PCA and SMOTE

Abstract

Liver is one of the most important parts of the body, liver is used for detoxification of medicine and toxin, filtering the blood, and many more. When liver is not functioning properly then the whole function of the body is going to be affected, resulting in someone to be in a dangerous situation and even death. An early diagnosis of the liver disease will let the patient to have an earlier treatment, therefore resulting in a higher chance of survival. This current is using supervised machine learning SVM and LR for the modeling, where in the preprocessing stage will be using the PCA and SMOTE to see the influenced of PCA and SMOTE in the machine learning model. Using Confusion Matrix as the performance evaluation, the prediction result of the model before PCA and SMOTE is LR with 70% of accuracy and SVM with 88% of accuracy, and after implementing PCA and SMOTE the accuracy result of LR is 64% of accuracy and SVM with 87% of accuracy. From the result, the model that gave a better prediction is SVM and both algorithm result has gone down while using the PCA and SMOTE method. But, seeing that the independent variable has been reduced to five, the change of the prediction result isn't that significant compared to the original dataset prediction result, and the prediction result have improved to a better prediction seeing from Precision, Recall, and F1-Score. From the aspect of execution time, the PCA and SMOTE method gave a better influence towards LR model than SVM.

Topics & Concepts