Frequently Asked Questions¶

Short answers to the questions students ask most. Each one links to the full material. If yours is not here, try the course schedule or ask on the course channel.

Where do I start?

Follow the learning path on the Home page top to bottom: Python → NumPy → Pandas → cleaning → EDA → visualization → Git → APIs → machine learning → Streamlit → PandasAI → LLMs → Ollama → agents → n8n → final project. The Course Schedule shows which week each step belongs to.

What is the difference between a session, a lab, and an assignment?

A session is a 90-minute lecture with notes and examples. The lab for that week (in Labs) is the hands-on exercise you submit. Assignments (1 and 2) are two larger graded projects that combine many sessions' skills.

Where are the quiz and exam answer keys?

They are intentionally not published on this site — quizzes and exams are graded in class and your instructor distributes the keys. See the Quizzes and Exams pages for format and how to prepare.

Where are the lab / assignment solutions?

Instructor solutions exist in the repository but are excluded from this website. Your instructor decides when to release them. Until then, the labs give you expected outputs so you can check your own work.

How is my final grade calculated? Can I track it myself?

The confirmed institutional scheme is Sessional 25 marks · Mid Exam 35 marks · Final Exam 40 marks (total 100). Because the maximum is exactly 100, Final Percentage = Sessional + Mid + Final — no normalization needed. The 25-mark Sessional is composed of Labs (8), Quizzes (5), Assignments (6) and the Final project (6). The letter grade and grade points follow the official scale in assessment plan §3. A ready-made Grade Calculator (Streamlit app) is included in the course repository: run streamlit run tools/grade_app.py.

Environment & Python¶

python is not recognized / not found

Python isn't on your PATH. On Windows re-run the installer and tick "Add python.exe to PATH"; on macOS/Linux try python3. Then open a new terminal. Full steps: Setup guide.

pip install fails or installs to the wrong place

You probably installed outside a virtual environment. Create one once per project: python -m venv .venv, activate it (.venv\Scripts\activate on Windows, source .venv/bin/activate elsewhere), then pip install -r requirements.txt.

Which Python / library versions does the course use?

Python 3.11+ with NumPy, Pandas, Matplotlib, Seaborn, scikit-learn, requests, jupyter (plus Streamlit for the apps module and pandasai/ollama for Module C). The pinned list is requirements.txt in the repo root.

Jupyter shows no kernel or won't start

From an activated venv run pip install jupyter, then jupyter lab. If a notebook still shows no kernel, pip install ipykernel and restart Jupyter.

Git & GitHub¶

How do I submit a lab?

Push your executed notebook (Run All, outputs visible) to your course repository and confirm the commit hash — the exact loop is in the Git & GitHub Guide.

git push says my changes were rejected

Someone pushed first. Run git pull, resolve any conflict the editor shows you, commit, and push again.

I committed the wrong file / a huge file

Unstage with git restore --staged <file>, add it to .gitignore, and make a new commit. For history rewriting (advanced) ask your instructor.

Streamlit¶

streamlit is not a recognized command

Your venv isn't activated, or Streamlit isn't installed there: pip install streamlit, then streamlit run app.py.

My app reruns the whole script when I click a widget — is that normal?

Yes — that is the Streamlit model. Use @st.cache_data for expensive loads and st.session_state to keep values between reruns. See the Streamlit module.

Charts don't appear / look broken in my app

Pass the figure explicitly: st.pyplot(fig) for Matplotlib, and use st.dataframe(df) (not st.write(df)) for large tables. The Streamlit FAQ covers the top gotchas.

PandasAI, Ollama & the AI tools¶

PandasAI says it can't connect or needs a model

PandasAI needs an LLM behind it. For this course, run Ollama first: ollama pull llama3.2, then create the agent with config={"llm": OllamaLLM(model="llama3.2")}. See the PandasAI Guide.

Ollama is slow or crashes on my laptop

Use a smaller model (llama3.2:3b), close other apps, and expect CPU-only inference to be slow — that is normal without a GPU.

Can I use ChatGPT instead of Ollama for the labs?

For labs graded on Module C you should use the local path so no data leaves your machine and the environment is identical for everyone. If your instructor allows a cloud model, never paste non-public data into it.

The AI gave me a number that is wrong. Did I do something wrong?

Possibly nothing — models hallucinate. The course rule is to verify one number by hand with pandas and log it. See the LLM & Ollama Guide and Session 30 · Ethics.

My agent loops forever / calls tools with nonsense arguments

Bound the loop (max_iterations), whitelist a few small tools, and validate arguments. The exact pattern is in the AI Agents Guide.

Do I need to install n8n for the n8n lab?

The lab works both ways: install n8n (npx n8n) for the visual part, or follow the workflow-in-Python pattern in Notebook 19 — see the n8n Guide.

Assessments & policies¶

When are the quizzes and exams?

Quiz 1: Week 8 Session 15 · Quiz 2: Week 15 Session 29 · Midterm: Week 8 Session 16 · Final: Week 16 Session 32. Full calendar on the Course Schedule.

Is using AI allowed in my submissions?

The course teaches AI-assisted work — but every AI-assisted step must be disclosed, verified, and logged, and quizzes/exams are AI-free. Read the policy in the Assessment Plan before you start anything.

How is the final project graded and presented?

Proposal → GitHub repo with README → report → 6-minute presentation in Week 16, plus an individual viva. Rubric and templates: Final Project.

Still stuck?¶

Streamlit FAQ — app-specific questions
Resources — official docs and free books
[GitHub Discussions / Issues] on the course repository — if you found a real bug in the materials, report it there