Aposemat IoT-23

Dataset Overview

Dataset Name and Source

• Dataset Name: Aposemat IoT‑23
• Dataset Source: iot_23_datasets_small
• Dataset Description: Contains labeled malicious and benign IoT network traffic created by Avast AIC Lab and funded by Avast Software.

Dataset Content

• Data Type: Labeled network traffic data
• Data Size: 8.8 GB
• Contents: Only includes labeled traffic data; PCAP files are not provided.

Data Processing and Analysis

Data Processing Stage

1. Data Cleaning and Pre‑processing

   • Jupyter Notebook: iot-23-data-cleaning-and-preprocessing.ipynb
   • Tasks include data loading, exploration, cleaning, processing, encoding, and storage.

2. Data Training

   • Jupyter Notebook: iot-23-data-training.ipynb
   • Trains and analyses multiple classification models, including Naive Bayes, K‑Nearest Neighbors, Decision Tree, Random Forest, LinearSVC, Artificial Neural Network (ANN), AdaBoost, and XGBoost.

3. Data Tuning

   • Hyper‑parameter tuning for the same set of models using GridSearchCV.

Data Storage

• Data Files: CSV-data
• Trained Models: applied-ML-DL-methods

Dataset Evaluation

• Evaluation Method: Stratified K‑Fold Cross‑Validator (StratifiedKFold)
• Parameters: 5 folds with shuffling enabled.