Comparative Analysis of Machine Learning Models for PDF Malware Detection: Evaluating Different Training and Testing Criteria

The proliferation of maliciously coded documents as file transfers increase has led to a rise in sophisticated attacks. Portable Document Format (PDF) files have emerged major attack vector for malware due their adaptability and wide usage. Detecting PDF is challenging its ability include various harmful elements such embedded scripts, exploits, malicious URLs. This paper presents comparative analysis machine learning (ML) techniques, including Naive Bayes (NB), K-Nearest Neighbor (KNN), Average One Dependency Estimator (A1DE), Random Forest (RF), Support Vector Machine (SVM) detection. study utilizes dataset obtained from the Canadian Institute Cyber-security employs different testing criteria, namely percentage splitting 10-fold cross-validation. performance techniques evaluated using F1-score, precision, recall, accuracy measures. results indicate that KNN outperforms other models, achieving an 99.8599% findings highlight effectiveness ML models accurately detecting provide insights developing robust systems protect against activities.