In this unit we introduce Python, Jupyter/Colab, and cover some fundamental concepts in Machine Learning, Probability and Linear Algebra.

1: Python Introduction

A fast tour of the Python you’ll actually use this year. Not a complete language course — we skip a lot — but enough that the rest of the units don’t trip you up on syntax.

What you’ll do

Work through the notebook cell by cell, running each one. Variables, data types, lists and dictionaries, control flow, functions, and imports. Then a first look at Matplotlib, which gets its own extended section at the end.

Key ideas

  • Cells run in the order you run them, not top to bottom — a common source of confusion later
  • Python is dynamically typed; you don’t declare types, but types still matter
  • Lists and dictionaries cover most of what you’ll need before pandas

2: Pandas and Real Weather Data

First contact with a real, imperfect dataset: Leesburg airport weather records. Real data has missing values, inconsistent labels, and columns whose names don’t tell you what they mean!

What you’ll do

Load a CSV into a DataFrame and start asking questions. How many rows? What are the columns? How many days were clear? Does it rain more on weekends?

Key ideas

  • A DataFrame is a table; almost everything is selecting rows, selecting columns, or grouping
  • df.describe() and df.info() before anything else — always look at your data before you analyze it
  • Missing data is normal and you have to decide what to do about it
  • The data format is described here
  • Notebook:

3: Weather Exercises

Eleven exercises in three tiers — Beginner, Intermediate, Advanced. Pick the tier that matches where you are.

What you’ll do

Answer questions about the weather dataset using pandas. The advanced tier includes a chi-square investigation into whether weekend weather is measurably different from weekday weather — your first hypothesis test, sneaking in during week two.

Open the exercises

Assignment: complete and submit your chosen tier.

4: Bayes’ Theorem

The probabilistic lens you’ll use all year. There are fundamental ways to measure success and failure, and knowledge and belief. We discuss them here.

Why it matters here

The classic medical-test problem: a test that’s 97% accurate, for a disease that affects 1 in 1000 people. Most people’s intuition about what a positive result means is badly wrong, and the gap between intuition and arithmetic is the whole lesson.

Notes (PDF)

5: London Weather Project

The unit’s capstone. A 50-year climate dataset and one question: has London’s weather gotten worse?

The task

You define “worse.” That’s the hard part and the point — part of the work is deciding what to measure, then defending that choice in writing. Temperature? Rainfall? Consecutive gray days? There’s no answer key.

What good work looks like

  • A clearly stated claim
  • Evidence from the data that actually supports it
  • Honest acknowledgment of what the data can’t tell you

Data: london_weather.csv (original source)

Assignment: submit your analysis notebook.

What’s Next

Unit 02 introduces the first actual model: linear regression derived from scratch, before any library does it for you.