Preprocessing and Analysis of an Open Dataset in Application Traffic Classification

Data preprocessing is a crucial step in data analysis and machine learning. This step involves transforming raw data into a suitable format for analysis, removing noise, and handling outliers to improve data quality. In particular, it offers benefits such as providing accurate analysis results, enhancing model performance, improving model generalization capabilities, and enabling faster model training. The public dataset (ISCX VPN-nonVPN 2016) is widely used in the field of application traffic classification. However, each study may have different methods for preprocessing the dataset, and detailed explanations or publicly available preprocessed datasets are not provided. Therefore, objective performance evaluation between methodologies becomes challenging. This paper performs preprocessing on the widely used public dataset (ISCX VPN-nonVPN 2016) in the field of application traffic classification and analyzes it. Additionally, it releases the preprocessed dataset publicly, enabling objective performance comparisons with other papers that utilize this public dataset.