Exercises for Data Science

General opinions

Get students doing something cool/motivational quickly - even if you need to give them training wheels
Use toy/reduced examples to introduce complex concepts and ideas (but only to introduce)
Use real data! It should have:
- stories/questions,
- column/variable names,
- be big enough to motivate using the tools you are teaching,

and sometimes it should be at least a bit messy.

Frequently ask students to interpret and communicate their results (open ended questions with critical manual grading!)

Derived from Teaching Tech Together by Greg Wilson

Code & Run - learner asked to write code that produces a specified output
Inverting Code & Run - learners asked to write tests to determine whether a piece of code conforms to a spec
Fill in the Blanks - learner is given some starter code and has to complete it so that the code produces a specified output
Parsons Problem - learners given lines of code that need to be rearranged to generate a specified output
Tracing
- execution: given a few lines of code, the learner has to trace the order in which those lines are executed
- values: given a few lines of code, the learner lists the values that one or more variables take on as the program runs
Minimal Fix - given a few lines of code that contain a bug, the learner must find it and make one small change to fix it
Refactoring - given a working piece of code, the learner has to modify it in some way without changing its output

Multiple choice questions - learners asked to choose the correct answer (works best with relevant distractors present)
Diagrams - provide them with a diagram and a set of labels and have them put the latter in the right places on the former
Matching - learner is asked to match labels in one column with text in another