DATA 3402 — Python for Data Science 2

Fall 2026 · University of Texas at Arlington · Instructor: Amir Farbin

Course materials are distributed through this repository. Pull regularly — lectures and labs are pushed as the semester progresses.

Contents

Path	What's there
syllabus.pdf	Course syllabus, grading breakdown, policies
Chapters/	The course text, one PDF per chapter, with a review question set
Lectures/	One folder per lecture: notebooks and/or slide PDFs
Labs/	One folder per lab assignment, plus the GitHub setup guide
sample.ipynb	Notebook whose one code cell prints Hello World!, to check Jupyter can run code
requirements.txt	Python packages used in the course

The course text

Chapters/ holds the written version of the material — the same ground the lectures cover, but in prose you can read at your own pace and go back to. Start with the chapter, use the lecture slides as the map, and use the review questions to find out what didn't stick.

File	Covers
Chapter.01.Computers.pdf	How computers work, from a transistor to an iPhone. Lecture 2 + Lab 1.
Chapter.01.Review.Questions.pdf	100 multiple-choice questions on Chapter 1, tagged easy / medium / hard.
Chapter.03.Probability.and.Statistics.pdf	Probability, Bayes, distributions, and generating data from them. Lectures 6 and 7.

The questions are for self-testing and are not graded. In most of them at least one wrong option is what you would believe if you almost understood the idea, so read them carefully.

Chapter 3 pairs with a notebook: read the chapter for the reasoning, then work through Lectures/Lecture.7/Lecture.7.ipynb, which turns it into code — a random number generator built from scratch, a histogram function written by hand, accept/reject sampling and Monte Carlo integration. Lab 3 draws directly on it.

More chapters land as the semester goes on.

Getting started

We did all of this together in the first lab session. These are the same steps, so you can catch up or start over.

You need Linux, macOS, or Windows with WSL. On Windows, WSL is required — not a suggestion. Google Colab is an acceptable fallback if your machine can't handle the later assignments, but it is not a substitute for having a working setup.

1. Get a Unix shell.

   • Windows: in PowerShell as Administrator, run wsl --install, then reboot. Launch Ubuntu from the Start menu and set a username and password when it asks. Write the password down — every sudo needs it.
   • macOS: open Terminal and run xcode-select --install.
   • Linux: you already have one.

2. Install the system packages. On Ubuntu/WSL/Debian:

   sudo apt update
   sudo apt install python3 python3-pip python3-venv git

   On macOS these came with the command line tools. Then close and reopen the terminal.

3. Make a course directory with its own virtual environment.

   cd ~
   mkdir -p Data-3402
   cd Data-3402
   python3 -m venv .venv
   source .venv/bin/activate

   Your prompt should now start with (.venv). That is a private Python for this course — nothing you install into it can break your system. Run the source line again in every new terminal; most "module not found" problems are just a forgotten activation.

   Do not use sudo pip install. Current Ubuntu blocks it with externally-managed-environment, on purpose; the virtual environment is the fix.

   Windows: keep all of this in your Linux home directory (~), not under /mnt/c. Git and Jupyter are slow and unreliable across that boundary. Your Windows drives are visible from Linux at /mnt/c, and your Linux files are visible from Windows File Explorer at \\wsl.localhost\Ubuntu.

4. Clone this repository and install the requirements. The clone is read-only; git pull brings you new lectures and labs as they are released. Submitting work also needs your own fork — see Labs and submission below.

   cd ~/Data-3402
   git clone https://github.com/UTA-DataScience/DATA3402.Fall.2026.git
   cd DATA3402.Fall.2026
   pip install -r requirements.txt

   No sudo on that last line — inside the virtual environment you are already allowed.

5. Check that it works.

   python3 -c "import numpy, pandas, matplotlib; print('ok')"
   jupyter lab

   Open sample.ipynb, click the code cell, and press Shift+Enter. Ctrl-C twice in the terminal stops Jupyter.

   Every time you come back: cd ~/Data-3402 && source .venv/bin/activate, then cd DATA3402.Fall.2026.

Notes on large data

Several lectures and labs use datasets too large to keep in git (the SUSY dataset, Kaggle competition data, image sets). They are not in this repository: you download them when a lecture or lab tells you to. .gitignore excludes them under the names the course uses, along with the pickles and model files the notebooks write — don't commit them. GitHub refuses files over 100 MB, so a committed dataset stops your pushes.

Labs and submission

Lab 5 has two parts. Labs/Lab.5/Lab.5.ipynb is the matrix class, the main assignment. Labs/Lab.5/Lab.5.CommandLine.ipynb is the shorter piece we work through in the lab session: getting a data file into shape from the command line, using bank-churn.csv beside it and files already in Lectures/. Nothing to download.

Lab work is submitted through your own fork of this repository:

• you fork this repository once on GitHub, giving you your own copy under your account;
• you pull from this repository to receive new lectures and labs as they are released;
• you do your lab work in your fork and push to your fork, which is where it is graded.

We set this up together in the Lab 4 session (Friday 9/11). If you missed it or need to start over, Part 1 of Labs/Lab.4/Lab.4.ipynb walks through it, and Labs/GitHub-Setup.pdf has the same steps line by line — including logging in to GitHub from the terminal, which you need before your first push — plus a table of what to do when git complains. In short:

1. Fork this repository. On its GitHub page, press Fork at the top right and fork it into your own account.

2. Clone your fork, not this repository:

   cd ~/Data-3402
   git clone https://github.com/YOUR_USERNAME/DATA3402.Fall.2026.git
   cd DATA3402.Fall.2026

   Already have the clone from Getting started? Then ~/Data-3402/DATA3402.Fall.2026 exists, and cloning your fork into the same place stops with destination path already exists. Don't clone again, and don't delete that folder: cd ~/Data-3402/DATA3402.Fall.2026 and carry on with step 3, which works on that clone unchanged.

3. Fix the remotes, so that pulls come from this repository and pushes go to your fork:

   git remote remove origin
   git remote add origin https://github.com/UTA-DataScience/DATA3402.Fall.2026.git
   git remote set-url --push origin https://github.com/YOUR_USERNAME/DATA3402.Fall.2026.git
   git remote -v

   git remote -v should show this repository for fetch and your fork for push. If both lines show the same URL, the set-url --push didn't take; run it again. If you use SSH keys, your fork's URL is git@github.com:YOUR_USERNAME/DATA3402.Fall.2026.git, in both the clone and the set-url --push.

From then on git pull brings class material and git push sends your work to your fork.

The moment you pull a new lab, copy it to a new name and work only in the copy — never edit the file we handed out:

cp Labs/Lab.3/Lab.3.ipynb Labs/Lab.3/Lab.3.solution.ipynb

git pull only touches files it recognizes from this repository, so a copy under a name we never handed out can't conflict with a fix we push later. To submit, commit your copy, push, and check on GitHub that your fork shows the new commit:

git add Labs/Lab.3/Lab.3.solution.ipynb
git commit -m "Lab 3 submission"
git push

Commit only your own copies. If you run a lecture notebook in place and commit it, the next release of that lecture can conflict when you pull; Labs/GitHub-Setup.pdf says how to recover.

Communication

All course communication goes through Teams — not email.