Comparison Analysis of Random Forest Classifier, Support Vector Machine, and Artificial Neural Network Performance in Multiclass Brain Tumor Classification

Abstract

This study aims to analyze and compare the performance of three main classification models, namely Random Forest Classifier, Support Vector Machine, and Artificial Neural Network, in classifying Multiclass brain tumors based on MRI images. The research method includes exploratory data analysis (EDA), dataset preprocessing with image segmentation using the Canny method, and feature extraction using the Humoment method. The performance of the classification models is evaluated based on accuracy, precision, recall, and F1 score. The analysis results show variations in the performance of the three classification models, with Random Forest Classifier having an accuracy of 0.7, weighted precision of 0.55, weighted recall of 0.7, and weighted F1 score of 0.59; Support Vector Machine having an accuracy of 0.71, weighted precision of 0.5, weighted recall of 0.71, and weighted F1 score of 0.59; and Artificial Neural Network having an accuracy of 0.62, weighted precision of 0.6, weighted recall of 0.62, and weighted F1 score of 0.61. Visualization using box plots also reveals outliers in the performance of the three models. These findings indicate variations and outliers in the performance of the classification models for Multiclass brain tumor classification. Further analysis is needed to understand the factors that influence performance differences and identify ways to improve the classification model performance for brain tumor diagnosis based on MRI images