How vduong421/ids-system works
Project Type and Purpose
• This is a data‐science/ML pipeline encapsulated in a single Jupyter Notebook (main.ipynb) for network intrusion detection.
• The notebook orchestrates data loading, preprocessing (including SMOTE resampling), model training (Random Forest, SVM, MLP), evaluation, and visualization.
High-Level Components
A. Data Ingestion
B. Data Cleaning & Preprocessing
C. Train-Test Splitting
D. Class Balancing (SMOTE)
E. Model Training (stock & tuned)
F. Evaluation & Metrics Calculation
G. Visualization & Reporting
Relationships & Data Flow
• Data flows sequentially from A → B → C → D → E → F → G.
• Within Model Training (E), three parallel tracks run (Random Forest, SVM, MLP), each with two configurations (stock vs tuned). Their outputs all feed into the Evaluation (F) block.
Architectural Patterns & Design Principles
• Pipeline Pattern: linear, stage-based processing of data.
• Parallel Model Training: fork/join pattern for training multiple models concurrently.
• Single workbook “host” that orchestrates all steps—there is no separate service or API layer.
Technology & Framework Call-Outs
• “Jupyter Notebook / Python” as the orchestration environment at the top.
• Under Data Ingestion (A), list “pandas” and “glob/os” for CSV merging.
• For SMOTE (D), list “imblearn.over_sampling.SMOTE.”
• In Model Training (E), list “scikit-learn” under each model.
• For Visualization (G), annotate “matplotlib” or “seaborn.”