Lesson 4 — Data Collection with APIs

Why this matters¶

Analytics and AI are only as good as the data pipeline behind them. A large part of professional work is:

• finding the right dataset,

• documenting what it is and where it came from,

• and ensuring you can reproduce the same results later.

Three ways to get data (concept map)¶

Today’s choice: OWID datasets (file-based, reproducible)¶

OWID offers high-quality public datasets with clear metadata and regular updates. We will treat OWID as a model for good practice:

• clear variable names,

• country and time identifiers,

• and a public data pipeline.

What to document (Data Source Note)¶

Mini case: a business/econ question using OWID¶

Example questions (choose one):

• “How did inflation evolve across countries after 2020, and how different are country experiences?”

• “How does GDP per capita relate to CO₂ emissions per capita (and does that differ by region)?”

• “How did vaccination rates and excess mortality vary across countries over time?”

Mini-lab (Google Colab)¶

In-class checkpoints

1. Load an OWID dataset into a DataFrame.

2. Filter for 3–5 countries and a time range.

3. Produce one plot and one summary table.

4. Write a Data Source Note and one limitation.

AI check (responsible use)¶

Review questions (quiz / reflection)¶

1. When is an API better than downloading a CSV?

2. What must be included in a Data Source Note for reproducibility?

3. Give one limitation of cross-country comparisons using public datasets.