Debrief after the course

[bast]

• we had too much content, too little depth (especially day 2)
• installation, setup, configuration: reserve a session 0 before we start
• also give people hints on how they can verify that they have working install
• add more exercises to day 1 and 3
• cover less cells in jupyter notebook
• increase frequency of exercises in day 2
• participants really wish to understand the content
• don't go into id(), however only show danger with copy with lists
• day 2 remove half the content, add smaller exercises
• session 5: not a success, either anonymous submission or replace session
• session 4: if something removed, remove pandas, but keep jupyter and matplotlib
• out with pandas, in with numpy
• with very few cells, show something convincing, add more exercises instead of just "shift enter"
• matplotlib examples: show how to print own concrete data instead of obscure numpy random data
• show f-strings but expect that some participants have python below 3.6
• editors: spyder was a nice find, but don't know how to pass CLI arguments on spyder
• explain better abstraction with functions, going from concrete to abstract, calling it with something concrete
• explain how things flow in and out of functions, variable naming inside and outside of functions
• confusion: variables vs. literals
• in teaching rather use f.open() and f.close() to avoid confusion about scope of with context

[bast]

Additional good suggestions I got:

• more exercises and a bit less talking
• divide the tasks in two or three different levels and construct them in a way such that it would be easy start them if you already know/get fast what the instructor is teaching