NumPy & Pandas for Beginners: A 20–22 Session Roadmap

By Mohit Agarwal, Paath.online•Published 4 August 2025•12 min read

Want to learn data analysis with Python without getting overwhelmed? This practical roadmap helps you build skills with the two essential libraries — NumPy and Pandas — across 20–22 guided sessions, with room to adjust practice time to your starting level and goals.

🗺️ Session-by-Session Plan (1–22)

1. Python refresher, notebooks, dataset setup
2. NumPy arrays: creation, shapes, dtypes
3. Indexing, slicing, boolean masks
4. Vectorization and broadcasting
5. Aggregation: sum, mean, axis operations
6. Pandas Series and DataFrame basics
7. Reading CSV/Excel, selecting rows/cols
8. Filtering, sorting, and type conversion
9. Handling missing values
10. GroupBy and aggregations
11. Merging and joining datasets
12. Reshaping: melt, pivot, stack/unstack
13. Datetime handling and time series basics
14. Apply/map, custom functions, vectorized ops
15. String and text data cleaning
16. Categories, sorting, and efficient data types
17. Combining datasets: practice with real-world joins
18. Exploratory data analysis and summary tables
19. Basic charts and communicating findings
20. Exporting results and making a repeatable notebook
21. Mini-project: clean and analyze a real dataset
22. Mini-project review, debugging, and next steps toward ML

⚠️ Common Pitfalls (and Fixes)

• Using pure Python loops in Pandas — prefer vectorization.
• Forgetting to set index or parse dates on load.
• Chained assignment warnings — use .loc with care.

🧩 Mini‑Project Ideas

• Analyze student scores across terms
• Create a simple sales report with trends
• Clean and explore a public dataset (e.g., Kaggle)

Study alongside the official NumPy and pandas manuals

A roadmap is only as good as the references behind it. Keep these tabs open while you practice so terminology stays precise and you learn stable APIs—not outdated blog snippets.

• NumPy fundamentals: start with NumPy quickstart and the ndarray reference.
• pandas workflows: follow the Getting started guides and keep the User guide handy for indexing rules and missing-data semantics.
• Python basics: when typing errors confuse you, cross-check docs.python.org for objects, iterators, and file I/O.

What comes after the roadmap: toward machine learning

Once you can clean a dataset end-to-end, you are ready to connect tables to models. Keep the first steps small and interpretable—linear models and decision trees beat opaque stacks for learning.

• Read scikit-learn's basic tutorial and work through one supervised learning example with a held-out test set.
• Browse educational notebooks and datasets on Hugging Face docs when you want modern deep-learning context—after you are solid on train/test evaluation.
• If you plan board-aligned AI electives, balance tool tutorials with policy reading on education.gov.in and notices on cbse.gov.in.