How nahla-nabil/ai-bootcamp---gdg-shaqra works
Project type and purpose
• This is a machine-learning classification pipeline focused on credit-card fraud detection.
• Implemented as a Jupyter Notebook (“ML Model - Nahla - AI Bootcamp (1).ipynb”) that ingests transaction data, preprocesses it, trains two MLP models (Scikit-learn and TensorFlow/Keras), and evaluates performance.
File structure
• Top level:
• A single notebook orchestrates the entire workflow.
• README describes purpose, architecture, data source, and tools.
• No separate “src” or “services” directories—everything happens inside the notebook.
README for additional insights
• Model architecture: Input (30 features) → Hidden layers (64 → 32 ReLU + dropout) → Sigmoid output
• Class imbalance handling: Oversampling
• Feature scaling: Standardization
• Technologies: Python 3, Pandas, NumPy, Scikit-learn, TensorFlow/Keras, Matplotlib, Seaborn