OPTIMASI RANDOM FOREST PADA KLASIFIKASI PENYAKIT LIVER BERDASARKAN PARAMETER KLINIS MENGGUNAKAN METODE GRID SEARCHCV

Abstract

This study aims to classify liver diseases based on patient clinical parameters using the Random Forest algorithm and improve model performance through hyperparameter optimization using the Grid SearchCV method. This test uses a dataset from Kaggle with 1,700 data points and 11 attributes. The research steps began with data exploration, data preprocessing such as data transformation and normalization, and the division of training and testing data into six trials, namely (60%:40%, 65%:35%, 70%:30%, 75%:25%, 80%:20%, 85%:15%). Next, hyperparameter optimization was performed using Grid SearchCV to find the most appropriate model settings. The optimization results showed an improvement in performance in all tests. The optimized Random Forest model produced the highest accuracy of 91.76% at a ratio of 85% training and 15% testing, as well as improving the precision, recall, and F1-Score values. Based on the test results, it can be concluded that the use of Grid SearchCV can improve the performance of Random Forest in classifying liver disease. This model can support the initial diagnosis process, thereby helping medical personnel identify patients who are potentially suffering from liver disease more quickly and accurately.