Data Science Foundations

This folder is the practical base layer for the rest of the repo. If you are not yet comfortable with arrays, DataFrames, plots, train/test splits, and the fit/predict/transform workflow, later LLM and deep learning phases will feel harder than they need to.

This is one of the highest-leverage phases in the repo. If this phase is weak, later notebook work turns into library imitation rather than actual understanding.

What This Phase Covers

• NumPy for numerical thinking
• pandas for messy real-world data
• matplotlib for basic visualization
• scikit-learn for classical ML workflows
• broader data science examples for exploratory and applied practice

Folder Map

• 1-numpy-examples/: array operations, broadcasting, indexing, exercises
• 2-pandas-examples/: cleaning, joins, grouping, time-series handling, projects
• 3-data-science-examples/: broader learning material and reference notebooks
• 4-matplotlib/: plotting fundamentals
• 5-scikit-learn/: a very large example library across major model families

Recommended First Pass

1. Work through 1-numpy-examples/
2. Move to 2-pandas-examples/
3. Use 4-matplotlib/ for core plotting habits
4. In 5-scikit-learn/, focus first on:
   • linear_model/
   • model_selection/
   • preprocessing/
   • ensemble/
   • cluster/
5. Use 3-data-science-examples/ as breadth and reinforcement, not as a strict sequential course

Study Advice

• Do not try to complete every scikit-learn notebook on the first pass.
• Prefer one full workflow over broad shallow browsing: data loading -> cleaning -> feature work -> split -> train -> evaluate -> explain results.
• Keep notes on leakage, validation mistakes, and metric choice. Those habits matter more than memorizing APIs.

How To Use This Phase Well

• Treat this as skill-building, not just exposure to libraries.
• Finish at least one end-to-end tabular workflow before moving on.
• Use the large scikit-learn folder selectively instead of trying to clear it like a checklist.
• Revisit this phase later when evaluation, leakage, or debugging issues show up in advanced projects.

Suggested Projects

• Iris or wine classification with proper validation
• Housing-price regression with feature engineering
• Customer segmentation with clustering and a short business write-up

What Comes Next

After this phase, move to:

1. 03-maths/README.md
2. 04-token/README.md
3. 05-embeddings/README.md

If you are still shaky on Python while doing this phase, loop back to 01-python/README.md alongside the exercises instead of waiting until you are stuck.