A Comparative Performance Analysis of Machine Learning and Deep Learning Algorithms for Web Attack Detection Using a GUI-Based Framework

Abstract

The rapid growth of web-based applications has significantly increased the vulnerability of systems to various cyber threats, including SQL injection, cross-site scripting (XSS), and denial-of-service (DoS) attacks. Traditional rule-based security systems are often inadequate in detecting evolving and sophisticated attack patterns. To address these challenges, this study presents a comprehensive performance analysis of multiple machine learning (ML) and deep learning (DL) algorithms for web attack detection using a graphical user interface (GUI)-based framework. The proposed system integrates five widely used classification algorithms: K-Nearest Neighbors (KNN), Random Forest (RF), Naive Bayes (NB), Logistic Regression (LR), and a Multi-Layer Perception (MLP) Neural Network. A user-friendly GUI is developed using Python’s Tkinter library to facilitate dataset loading, model execution, and visualization of results. This interface enables users to easily compare the performance of different algorithms without requiring deep technical expertise. The system processes datasets in CSV format, applying preprocessing techniques such as label encoding for categorical variables and feature scaling using standardization. The dataset is then split into training and testing subsets to ensure unbiased evaluation. Each algorithm is trained and evaluated using key performance metrics including Accuracy, Precision, Recall, F1-Score, and Area Under the Curve (AUC). The results are displayed in a tabular format and visualized through bar charts for intuitive comparison.Experimental results demonstrate that ensemble methods like Random Forest generally outperform other models in terms of accuracy and robustness, while neural networks provide competitive performance with the ability to capture complex patterns. Simpler models like Naive Bayes and Logistic Regression offer faster computation but may lag in handling non-linear relationships in data. The main contribution of this work lies in providing an interactive platform for evaluating multiple algorithms under a unified framework, making it highly suitable for educational purposes, research experimentation, and preliminary cyber security analysis. The GUIbased approach enhances usability and accessibility, allowing even non-expert users to perform advanced machine learning analysis. In conclusion, this system highlights the importance of selecting appropriate algorithms for web attack detection and demonstrates the effectiveness of machine learning techniques in enhancing cyber security measures. Future work may involve integrating real-time data streams, advanced deep learning architectures, and automated hyper parameter tuning to further improve detection performance.