How qwerty30112/2nd-ml100days works
Project type and purpose
• This repository (“2nd-ML100Days”) is a collection of daily Jupyter notebooks, each corresponding to one day’s homework in a 100-days machine-learning challenge.
• Its purpose is educational: to step through data loading, preprocessing, model training, evaluation, and visualization tasks day by day.
File structure
• Top-level files:
• Day_001_HW.ipynb … Day_017_HW.ipynb – Jupyter notebooks for each day
• README.md – project title only
• No separate “data/”, “src/”, or “models/” folders, so each notebook is self-contained.
• Likely reliance on external data files or public datasets loaded inside the notebooks.
README for additional insights
• README only contains the project name; no technology or architecture hints.
a. Main components
• Data Sources (CSV / public datasets)
• Jupyter Notebook Environment
• Preprocessing Modules (code cells in notebooks)
• Model Training & Hyperparameter Tuning
• Model Artifacts (saved models / checkpoints)
• Evaluation & Visualization Components
• Compute Environment (local machine, Colab, or cloud VM)
b. Component relationships & interactions
• Data Sources feed into Preprocessing cells
• Preprocessing output feeds Model Training
• Model Training produces Model Artifacts
• Model Artifacts plus Test Data feed Evaluation & Visualization
• All notebooks run in the Compute Environment, which reads/writes to a shared workspace (e.g., local disk or cloud storage)
c. Architectural patterns & design principles
• Linear pipeline per notebook (Data → Process → Model → Evaluate)
• Reusable code cells (can be an implied “utilities” layer)
• Stateless notebooks, but with persistent artifacts (models, figures)
d. Key technologies & libraries
• Jupyter Notebook (front-end interface)
• Python data stack: pandas, NumPy
• ML frameworks: scikit-learn, TensorFlow or PyTorch (depending on day)
• Visualization: Matplotlib, Seaborn
• Storage: local filesystem or cloud bucket