Pandas

Beginner → Intermediate · 6 topics

pandas works with tables — rows and named columns, like a spreadsheet you control with code. In GenAI work you'll use it to load documents, clean data, prepare RAG chunks, compare models and turn LLM logs into reports. It's built on NumPy.

pip install pandas          # or: uv add pandas

All the pages use one small, realistic dataset — LLM evaluation results:

model	question	latency_ms	tokens	cost_usd	rating
gpt-4o-mini	What is RAG?	820	210	0.00013	5
llama-3.1-8b	What is RAG?	410	190	0.00002	4
…	…	…	…	…	…

#	Topic	Sub-topics
1	Series & DataFrames	Create tables · read CSV/JSON · inspect: head, info, describe · save
2	Selecting & filtering	Columns · loc / iloc · conditions · isin, str.contains · query · sorting
3	Cleaning data	Missing values · types · dates · duplicates · text cleanup · new columns
4	Grouping & aggregation	groupby · named aggregations · transform · value_counts · pivot tables
5	Combining & reshaping	merge (joins) · concat · melt · pivot · explode
6	Pandas for GenAI	RAG chunks · token & cost reports · model comparison · JSONL for fine-tuning

Pandas thinks in columns

Write df["cost"] * 1000, not a loop over rows. Column operations are short, readable and run at NumPy speed.

Next: Series & DataFrames →