ECON 8310 — Homework Quickstart

Start Here

Which repository am I looking at?

There are two ECON 8310 repositories, and you only ever clone one of them.

Repository	What it holds	What you do with it
`forecasting-env` — this one	The assignments, the datasets, and the folder your work goes into	Clone it. Every homework you write and submit lives here
`forecasting-course`	Slides, labs, assignments, the book and the workbook — the teaching materials	Nothing. You read these on the web, not in a clone

Course website (slides, labs, assignments, book, workbook): https://www.luozijun.com/forecasting-course/

Your syllabus, the term calendar and lecture recordings are on Canvas, not on the website.

If you are looking for a lecture or a lab, it is on the website. If you are looking for homework, it is here. Everything you need for ECON 8310 homework is in this repository.

1. One-time setup

Get this repository onto your machine.

git clone https://github.com/DataHurdler/forecasting-env.git

Or open https://github.com/DataHurdler/forecasting-env, click the green Code button, and download the ZIP — then unzip it somewhere you will find again.

Set up Python. You need Python 3.10 or newer and these packages:

jupyter pandas numpy matplotlib statsmodels scikit-learn
prophet pygam xgboost torch pymc arviz

jupyter is not optional — Quarto runs your Python through a Jupyter kernel, and rendering fails without it even though nothing in your code imports it.

macOS only, before Week 5: brew install libomp, once. Without it import xgboost fails.

Install Quarto from https://quarto.org/docs/get-started/, and an editor — either VS Code (with the Quarto extension) or Positron (Quarto built in). Either works.

Full step-by-step setup, what a .qmd file actually is, and the four errors you will hit: QUARTO_GUIDE.md. Read that first if you have not used Quarto before.

Google Colab will not work — it cannot open .qmd files.

When something breaks, read Troubleshooting — it covers the failures this course actually produces, and it is faster than waiting on a reply.

2. Get the data

Every assignment reads from data/processed/. Try building it yourself first — it is worth seeing where the data comes from:

python scripts/prep_m5.py            # the main dataset
python scripts/prep_fred.py          # macro series, Lecture 2
python scripts/prep_electricity.py   # hourly demand, Lecture 3

prep_m5.py downloads about 48 MB and takes a few minutes.

If that fails for any reason, you are not stuck. The finished files are already committed in data/processed/. Nothing in this course is blocked by a download.

3. Do an assignment

Copy the assignment out of assignments/ into a new folder:
```
submissions/hw01_part1_<yourname>/HW01_Part1_ETS.qmd
```
Use the exact assignment name for the folder prefix — hw01_part1, hw05_part2, and so on. It is written at the top of each assignment.
Open the assignment and find Initial Prompt for Your AI Assistant. Send that block as the first message to whichever AI assistant you are using. Save it as INITIAL_PROMPT.md in your folder.
Work through the questions. Fill in the code chunks and write the answers.
Keep a PROMPT_LOG.md in your folder as you go (see below).

Render it:

quarto render submissions/hw01_part1_<yourname>/HW01_Part1_ETS.qmd

4. Check before you submit

python scripts/check_my_submission.py

No arguments. It tells you what is missing. The most common way to lose marks is submitting an .html that is older than your .qmd — you edited, then forgot to re-render. This catches that.

5. Submit

Submit on Canvas. Open the assignment and upload the contents of your submission folder:

File
`HWxx.html`	the rendered document — self-contained, so this one file carries all your code, output and figures
`HWxx.qmd`	the source
`PROMPT_LOG.md`	every prompt you sent, numbered
`INITIAL_PROMPT.md`	the context prompt you started with
`REPORT.md`	only where the assignment asks for one

Attach them individually, or zip the folder and upload that. Either is fine.

Upload before the deadline. The Canvas timestamp is submission time.

You never push to this repository

This repository is your working folder, not your submission. It is mine, and you do not have write access — git push will fail, and that is expected, not something you have broken.

Two things follow:

git pull is useful. If an assignment is corrected during the term, pulling brings you the fix without re-downloading anything.
Committing locally still works. A clone with no remote is a perfectly good local history. Commit as often as you like; it costs nothing and it is a good habit.

Keeping your own copy on GitHub is optional and entirely up to you. If you do it, make the repository private — a fork of a public repository is always public, and a public copy would put your work and your prompt log in front of the whole class.