Machine Learning (and analysis) for Public Interest Technologists

A Jupyter Book which contains learning modules for teaching applied machine learning and analysis for public interest technologists. This course specifically focuses on transcripts and "text-as-data."

Assumptions / Prerequisites

1. You have experience with Python and common scientific Python libraries (pandas, seaborn, etc.).

Learning Objectives

1. Gain a basic understanding of the data produced by municipal governments, especially in the legislative branch.
2. Understand processing methods to turn "messy" text into data ready for statistical analysis.
3. Understand data annotation practices and methods.

Acknowledgements

This repository was inspired by Brian Keegan's Web Data Scraping Course.

Examples from Others

General Notes

• always link to others work to reduce our own work

Brian Keegans Course

• Exercises (hook up to binder)
• Explain the data model(s) in detail
• Explain the functions being used

Supervised ML

• FIRST CHAPTER IS "WHAT YOU WILL LEARN / DO"
• This course is more theory -> case studies of real use. ML-for-PIT is direct to application?
• Lots of background info in first chapter alone.

Generally the structure is:

• Include some summary of the chapter goals.
• Try to include mid-point questions.
• Run some code
  • fill in some parameters / constants
• you were able to....
• but wait, there may be problems
• now fix those problems

Regression Analysis

• What is ML

• What is PIT

• I like the FAQ model

• course objectives

Generally the structure is:

• summary of chapter goals
• Show code to do things
• Explain theory behind code
• Detail / follow the line of thinking
• Recap