Analisis SMOTE Pada Klasifikasi Hepatitis C Berbasis Random Forest dan Naïve Bayes

Abstract

According to WHO, around 71 million people were infected with the Hepatitis C virus in 2019. However, only 49.7% of people are aware of Hepatitis C. Early prevention is essential to minimize the possibility of something terrible. To maximize the efforts of medical experts in minimizing the risk of transmission, a program was created that is capable of classifying Hepatitis C with an automatic detection system using a machine learning model. Random Forest was chosen because it can handle outlier and imbalance data so that it can produce high accuracy values and can identify important features. Naïve Bayes was chosen because of its simple algorithm, but capable of producing high-accuracy values. After testing both models, the confusion matrix formula calculates the prediction results. The test results show that applying the Random Forest model without SMOTE is 93%, and Naïve Bayes without SMOTE is 88%. Due to the data imbalance in the dataset, an oversampling technique is performed using the SMOTE method. The test results were obtained by applying the Random Forest model with a SMOTE of 98% and Naïve Bayes with a SMOTE of 89%.

Classification of Sleep Disorders Using Random Forest on Sleep Health and Lifestyle Dataset

“…This model uses a combination of multiple decision trees to make predictions. The steps include determining the number of trees, building decision trees with random subsets of training data, making predictions by combining the prediction results from each tree, evaluating the model with appropriate evaluation metrics, and saving the model for future use [9], [10]. Random Forest can overcome overfitting, provide stable and accurate predictions, and provide information about the importance of features in prediction.…”