Pandas: slim down / restructure the lesson

[ashwinvis]

It was my first time teaching this, so I will be try to be as constructively critical as possible. In my opinion:

The good parts

• Good example data: titanic, nobel prize
• Explanation with code to type-along
• Good practice advice like tidy data
• Exercise 1: is about selecting and indexing.
• Exercise 3: contains a repetition for computing age_nobel (and that is a good thing) based on the calculation for lifespan

The bad parts

• Lacks structure. For example:
  • Some topics like groupby are introduced too early and are a bit too complicated
  • A potpourri of methods are thrown in. The 10 minutes to pandas tutorial and the cheatsheet can potentially give ideas to regroup the sections in the episode. Progressively increasing the difficulty is the key.
• Section headings "working with dataframes" and "time series superpowers" are too vague and doesn't help the narrative.
• Exercises too long / lacks clarity on what the learning outcome is.
  • Important things like dealing with missing data does not have exercises.
  • Exercise 1: first two tasks are too open-ended. It also has the disctraction on missing data just before it.
  • Exercise 2
    : purpose unclear after talking about concat, merge and groupby
  • Exercises 3: some of the basic steps are OK, but they appear detached from the narrative.

Possible additions

• Lots of question on the notes comparing to R and SQL. This can be hinted early on.
• Add computation of speed = distance / time in the Tidy data discussion to make a point on "Tidy Data" clear.
• Convert some exercise questions into homework, so it is clear what is expected from the learners during the course