"\"CIC-DDoS2019 Non-Scaled Balanced 8 Attack Subset\""

"The CIC-DDoS2019 dataset was initially released by the Canadian Institute for Cybersecurity. On the testing day, seven DDoS attacks were executed, while twelve occurred on the training day to compile this dataset. The uploaded data titled \"CIC-DDoS2019 Non-Scaled Balanced 8 Attack Subset\" is prepared by applying a series of modification steps on the training day traffic of CIC-DDoS2019 dataset.  These steps encompass:Data integration- Data is integrated from 12 CSV files which are generated by the network traffic of training day.Preprocessing:Removal of spaces in column names.Removal of 10 non-informative columns: 'Unnamed:0', \"FlowID\", \"SourceIP\", \"DestinationIP\", 'FwdAvgBytes\/Bulk', 'FINFlagCount', 'FwdAvgBulkRate', 'PSHFlagCount', 'Timestamp', 'SimillarHTTP'.Removal of rows with infinite valuesRemoval of  duplicate rowsSelection of a subset of data comprising  benign traffic along with eight specific attacks: NTP, DNS, LDAP, MSSQL, NetBIOS, SNMP, SSDP, UDP, UDP-Lag, WebDDoS, SYN, and TFTP.Drop the highly correlated 30 features from the Dataset. Number of features after dropping highly correlated ones are 47.Dataset with 47 features are down sampled with random under sampling to produce the final balanced dataset."

CIC-DDoS2019数据集（CIC-DDoS2019 dataset）最初由加拿大网络安全研究所（Canadian Institute for Cybersecurity）发布。在该数据集的构建流程中，测试阶段共实施7类DDoS攻击（Distributed Denial of Service Attack），训练阶段则开展12类攻击以完成数据集的编译。本次上传的命名为‘CIC-DDoS2019非缩放均衡8攻击子集’（CIC-DDoS2019 Non-Scaled Balanced 8 Attack Subset）的数据，是基于CIC-DDoS2019数据集的训练阶段网络流量，经过一系列处理步骤生成的。这些处理步骤具体包括：
1. 数据整合：从训练阶段生成的12个CSV文件中整合原始数据；
2. 预处理环节：
- 移除列名中的空格字符；
- 删除10个无信息价值的列：'Unnamed:0'、'FlowID'、'SourceIP'、'DestinationIP'、'FwdAvgBytes/Bulk'、'FINFlagCount'、'FwdAvgBulkRate'、'PSHFlagCount'、'Timestamp'、'SimillarHTTP'；
- 删除包含无穷值的样本行；
- 移除重复样本行；
- 选取包含良性流量与8类特定攻击的数据集子集，涉及的攻击类型包括：NTP、DNS、LDAP、MSSQL、NetBIOS、SNMP、SSDP、UDP、UDP-Lag、WebDDoS、SYN及TFTP；
- 移除数据集中高度相关的30个特征，移除后剩余47个特征；
- 对包含47个特征的数据集采用随机欠采样方法进行降采样，以得到最终的均衡数据集。