Data Driven Modeling - Day 1
Repository: https://github.com/colbrydi/data_driven_modeling
Agenda (80 minutes)
- 10 min - Check-in, introduction to the data-driven-modeling unit, and start building the Conda environment
- 15 min - Finding a dataset and how web APIs work
- 40 min - Guided work in
Assignment_Instructions_Part_1.ipynb - 15 min - Discuss what summaries and trends can and cannot claim
Learning Goals
By the end of this class, you should be able to:
- Request data from a web API with Python and record where the data came from.
- Summarize a time series and explain what those statistics cannot tell you.
In-Class Focus
This class begins the data-driven-modeling unit with the simplest data-driven model: statistics. We start with real weather data from the Open-Meteo web API, compare it with an API that requires a key, and then ask what means, confidence intervals, and trends can support before any machine learning is added.
Use the repository as the primary working reference:
- Repository: https://github.com/colbrydi/data_driven_modeling
- Notebook:
Assignment_Instructions_Part_1.ipynb - Setup:
conda env create --prefix ./envs --file environment.yml
Start the environment build during check-in because it can take several minutes. As you work, consider where the data came from, whether someone else could repeat your download, and which assumptions (independent observations, a stable baseline) limit your conclusions.
Before Next Class
- Finish the remaining sections of
Assignment_Instructions_Part_1.ipynb, including at least one “Make it yours” option. - Write down one limitation of your analysis that a summary statistic does not show.
- Preview
Assignment_Instructions_Part_2.ipynb.