DSCI 100 Course Project: Set Up

📋 Your course project

📋 Project: what you hand in

• The project will be self-contained in one Jupyter notebook. Fill in the cells with your code and your text.
• An individual project: your own work, your own repository.
• Reproducible: it must run from top to bottom without errors, by itself, with no other files.
• You are required to use the methods you learned from DSCI 100. If you look something up elsewhere (e.g. how to handle missing data), put a link at the end of your report. A blog post or StackExchange link is fine.

• Callback to the recipe analogy for reproducibility: someone else should be able to follow your notebook and get the same result.

📋 Project: checkpoints

Checkpoint	Done up to	Due (11:59pm)
A	Task 0.1: your name, repository link and seed; and upload your repository URL to Canvas	Saturday, Oct 10
B	Task 3.1	Saturday, Oct 31
C	Task 5.1	Saturday, Nov 7
Final	The whole report, .ipynb and .html, also uploaded to Canvas	Saturday, Dec 5

• A checkpoint is a draft. You can keep changing your work after it.
• We will look at what is on your GitHub repository right after 11:59pm. A commit you never pushed does not count.

• The points here are only for handing in on time. The other 74 points are for the report itself.
• I take a snapshot of every repository right after each deadline, so pushes after 11:59pm are not in it.

📋 Project: Generative AI

• You have all the tools you need to write this code yourself.
• ✅ Use AI as an advanced search engine, e.g. to understand an error.
• ❌ Don’t use it to write the code or text in your report.
• Any AI use must be disclosed at the bottom of your report.

Example of a good disclosure: “I tried to fit a model with function X and got error Y. AI helped me see the error came from a missing value.”

Important

Not disclosing AI use is academic misconduct. Submissions suspected of inappropriate use of AI will be flagged by the teaching team, and handeled by a case-by-case basis.

Project assignment details are on Canvas

Set up your project step by step

Tip

If you prefer a video walkthrough, you can watch this video.

🔬 Demo: set up your project with me

Follow along on your laptop. Every step stays on this page, so you can come back to it.

Part	Where	What you do	Git word
1	GitHub	Make your own copy of the project template
2	JupyterHub	Copy your repository into JupyterHub	clone
3	JupyterHub	Fill in your name, link and seed
4	JupyterHub	Take a snapshot	add, commit
5	JupyterHub → GitHub	Send the snapshot to GitHub, with a token	push
6	GitHub → JupyterHub	Bring a change made on GitHub back	pull
7	JupyterHub → GitHub	Add a new file of your own	add, commit, push

Part 1: Make your own repository

📋 Step 1: Use the template

Open the project template. Click Use this template, then Create a new repository.

• Not Fork. A fork stays linked to my repository, and GitHub lists every fork on it, so anyone could click through to everyone else’s project.
• Make sure they are logged in to GitHub first, or the button is missing.

📋 Step 2: Name it and keep it public

Name it dsci-100-project- plus your first and last name, e.g. dsci-100-project-sky-sheng. Leave it Public. Click Create repository.

• Public is required, or we cannot grade it.

📋 Step 3: Copy your repository’s address

Click the green Code button. Under HTTPS, click the copy icon. You need this address three times: to clone, in your notebook, and on Canvas.

Part 2: Copy your repository into JupyterHub

📋 Step 4: Open JupyterHub from Canvas

On Canvas, click JupyterDSCI100 in the menu on the left.

📋 Step 5: Start your server

Click Start My Server. It can take a minute.

📋 Step 6: Stay in the top folder

You should see / at the top and the dsci-100-student folder below it. Stay here. Don’t open dsci-100-student.

📋 Step 7: Open the Git tab

Click the Git icon (the diamond) on the far left, then Clone a Repository.

📋 Step 8: Paste your address and clone

Paste the HTTPS address from step 3. Leave both boxes as they are. Click Clone.

📋 Step 9: Your repository is here

You see Successfully cloned, and a folder with your repository’s name appears on the left.

Part 3: Fill in your details

📋 Step 10: Open your notebook

Double click your repository’s folder, then double click dsci-100-project-notebook.ipynb.

📋 Step 11: Edit the name cell

Scroll to Part 0. Double click the cell with First and Last Name to edit it.

📋 Step 12: Name, link and seed

Fill in your name and your repository’s HTTPS address from step 3. Change the seed to your student number. Run both cells, then save.

⚠️ Save with Cmd+S (Mac) or Ctrl+S (Windows). Git only sees what is saved.

Part 4: Take a snapshot

📋 Step 13: Choose what goes in the snapshot

Click the Git icon again. Hover over your notebook under Changed and click its +.

• Staged: what goes in your next snapshot.
• Changed: files Git already knows, that you edited.
• Untracked: new files Git has never seen. These two are backup copies JupyterHub made. Leave them alone. You’ll add a file of your own in step 40.

• The + is the add step from the lecture: it lines the file up for the next snapshot.
• Don’t click the + on the Untracked heading. It adds every untracked file, backups included, and those have no place in the repository.

📋 Step 14: Write a message and commit

Your notebook is now under Staged. Write a short message about what you did, then click Commit.

• Good message: what changed, in a few words. “Fill in my name and GitHub repository link”, not “update” or “asdf”.

📋 Step 15: Tell Git who you are

First time only. Type your name and the email you used to make your GitHub account. Click Ok.

Git writes your name and email on every snapshot. GitHub uses the email to match the snapshot to your account.

• Any email works for the commit itself. With the email on their GitHub account, their username and picture show next to the commit on GitHub. With another email, the commit still goes through, but it shows a plain name with no link to their account.
• If they set their GitHub email to private, push can fail with GH007. See the FAQ slide.

📋 Step 16: Check GitHub

Your repository on GitHub still shows 1 Commit. Open the notebook there: your name is not filled in yet.

Commit saves the snapshot in JupyterHub only. GitHub has not heard about it yet.

Part 5: Send it to GitHub

📋 Step 17: Push

Back in JupyterHub, click the cloud with the up arrow. That is push. The orange dot means you have commits GitHub doesn’t have yet.

📋 Step 18: GitHub asks who you are

A window asks for two things.

Important

• username: your GitHub username, e.g. skysheng7. Not your real name.
• personal access token: not your GitHub password. You don’t have one yet, so click Cancel and make one next.

Part 6: Make a personal access token

🔑 GitHub wants a token, not your password

• A personal access token is like a hotel key card: it opens only the doors you allow, it expires, and you can cancel it at any time.
• Your password is the master key to your whole account. GitHub won’t take it from other apps like JupyterHub.
• You make the token once, and use it every time you push.

🔑 Step 19: Open your settings

On GitHub, click your profile picture in the top right, then Settings.

🔑 Step 20: Developer settings

Scroll to the bottom of the menu on the left. Click Developer settings.

🔑 Step 21: Generate a classic token

Click Personal access tokens, then Tokens (classic). Click Generate new token, then Generate new token (classic).

Curious: classic or fine-grained?

A fine-grained token is newer and can be limited to one repository. That is safer, but you have to pick the repository every time, and a new repository needs a new token. A classic token works for all your repositories, which is simpler for this course.

🔑 Step 22: Name it and set an expiry

Under Note, give it a name you will recognize, e.g. DSCI 100. Set Expiration to 90 days, so it lasts past the final deadline.

🔑 Step 23: Tick repo only

Tick the box next to repo. Its sub-boxes tick themselves. Leave every other box empty.

Why only repo?

Each box is one more door the key card opens. Pushing to your repository only needs repo. The other boxes are for things like running automated workflows, managing organizations, or deleting repositories, which this course never uses. If a token with every box ticked ever leaks, someone could delete all your repositories with it.

🔑 Step 24: Generate the token

Scroll to the bottom and click Generate token.

🔑 Step 25: Copy it now

Click the copy icon next to your token, and save it somewhere safe, like a password manager or a note on your own laptop.

Important

• GitHub shows the token only once. If you lose it, generate a new one and delete the old one.
• Never share it with anyone, and never paste it in your notebook. Your repository is public, so the whole world would see it.

Part 7: Push again

📋 Step 26: Push with your token

Back in JupyterHub, click push again. Type your GitHub username, paste your token, tick Save my login temporarily, and click Ok.

With the box ticked, JupyterHub remembers your token for a while, so you don’t paste it on every push. When it asks again, paste the same token.

• “Temporarily” is real: the JupyterHub Git extension keeps the login in memory for about an hour by default, and it is gone after a server restart. So they will be asked again on another day. That is why step 25 says to save the token somewhere safe.
• Only tick it on their own JupyterHub account, which is the only one they use.

📋 Step 27: Pushed

You see Successfully pushed.

📋 Step 28: Check GitHub again

Refresh your repository on GitHub. It now shows 2 Commits, with your message at the top.

The time shown is when you committed in JupyterHub, not when you pushed.

📋 Step 29: Your changes are on GitHub

Click the notebook on GitHub. Your name, link and seed are there.

That is Checkpoint A done. 🎉

Part 8: Bring changes back with pull

📋 Step 30: Edit a file on GitHub

Sometimes you change a file on GitHub instead of JupyterHub. Let’s try it. On your repository, click the pencil next to README.

📋 Step 31: Add a sentence

Type a new sentence at the bottom. Click Commit changes….

📋 Step 32: Commit on GitHub

Write a commit message, then click Commit changes. This snapshot is made on GitHub, so there is nothing to push.

📋 Step 33: JupyterHub doesn’t know yet

Open README.md in JupyterHub. Your new sentence is not there.

• Same idea as step 16, the other way round. GitHub and JupyterHub only sync when you push or pull.

📋 Step 34: Pull

In the Git tab, the cloud with the down arrow has an orange dot: GitHub has a snapshot you don’t. Click it. That is pull.

📋 Step 35: Pulled

You see Successfully pulled.

📋 Step 36: In sync again

Open README.md again. Your sentence is there.

⚠️ If you ever edit on GitHub, pull in JupyterHub before you start working. Otherwise your next push gets rejected.

Part 9: Add a new file of your own

📋 Step 37: Make a new file

Check you are inside your repository’s folder (see the path at the top). Right click on the empty space below your files, then New File.

• Only files inside the repository folder can go to GitHub. A file made in the top folder is outside the repository, and Git never sees it.

📋 Step 38: Name it

Type a name, e.g. my_note.txt, and press Enter. Double click it to open it.

• Keep the .txt ending, so JupyterHub opens it as plain text.

📋 Step 39: Write something and save

Type a sentence. Then File → Save Text, or Cmd+S (Mac) / Ctrl+S (Windows).

• Same rule as step 12: Git only sees what is saved. The dot on the tab means unsaved changes.

📋 Step 40: Add only your file

Open the Git tab. my_note.txt is under Untracked: Git has never seen it before. Hover over it and click its own +.

• ✅ The + next to one file adds just that file.
• ❌ The + on the Untracked heading adds every untracked file, including JupyterHub’s backups.

The files with checkpoint in their name are backups JupyterHub saves on its own, so you can go back to an earlier version. They are not part of your project, so they stay untracked.

• Untracked does not mean bad. It only means “Git has never had this file in a snapshot”. Every new file starts there.
• The .ipynb_ch... next to some names is the hidden .ipynb_checkpoints folder they live in.
• If someone already added the backups: it is not a disaster, the project still works. They can be removed later in office hours.

📋 Step 41: Commit it

my_note.txt is now under Staged. The A next to it means a new file is being added.

Write a message and click Commit.

📋 Step 42: Push

The push button has an orange dot again: 1 commit GitHub doesn’t have. Click it.

You ticked Save my login temporarily in step 26, so JupyterHub pushes without asking. Refresh your repository on GitHub: my_note.txt is there.

• If it has been more than about an hour, or the server restarted, the sign in window comes back. Same as step 26: GitHub username, the same token, tick the box, Ok.

Hand in your work

📋 Submit your repository link on Canvas

• On Canvas, open the assignment Project: GitHub Repository Link.
• Paste your repository’s HTTPS address from step 3, and submit.
• Due with Checkpoint A, Saturday, Oct 10, 11:59pm.
• We use this link to grade every checkpoint, so make sure it is right, and your repository is public.

• Quick check they can do: open the link in a private browser window. If they see their repository without logging in, it is public and the link works.

📋 Every checkpoint: the same loop

1. Save your notebook.
2. Git tab: click + next to your notebook to stage it. Never the + on the Untracked heading.
3. Write a message, click Commit.
4. Click push (cloud, up arrow).
5. Open your repository on GitHub and check your commit is there.

If it is not on GitHub, we can’t see it.

🆘 FAQ: common errors

You see	What it means	Fix
Authentication failed, or the sign in box keeps coming back	Wrong username, or a password instead of a token	GitHub username, then paste the token. Expired? Make a new one.
Repository not found when cloning	Typo in the address, or the repository is private	Copy the address again from the green Code button. Check it is public.
Updates were rejected because the remote contains work that you do not have locally	GitHub has a snapshot you don’t, e.g. an edit made on GitHub	Pull, then push.
nothing to commit, working tree clean	Nothing new to save	Save your notebook first, then stage it with +.

❓ Questions?

• Ask on Piazza, or come to office hours. Bring a screenshot of the error.
• Want to save these slides? Right click → Print → Save to PDF.