Lesson selection and customization.

[bsmith89]

I prefer the structure of gapminder over inflammation, but I think the biggest differences will be on Morning 1 so whoever takes the that half should have the final say.

In any case, we won't be teaching the shell, so we probably don't have a way to teach running Python from the CL (unless we want to take some time to teach shell stuff, which is an option).

I'm also curious if we want to focus on data analysis libraries (pandas, matplotlib, statsmodels, seaborn, numpy etc.) more than the base lesson (maybe more in the DC style?). Jackie suggested proposing a syllabus for Morning 2, but taking feedback on that plan at the end of the first morning.

[aerenchyma]

I agree that I prefer gapminder over inflammation. Assuming I take Morning 1, I am definitely fine with doing that.

I also agree that teaching the shell/REPL for a novice class is confusing and focusing on programs are files, how to run the files, what happens if you mess up and just type "python" (how to exit the REPL, more on that later) is a good idea.

WRT focusing on data analysis libraries, I think that might be good to have a proposed syllabus in part so we can get a sense on morning 1 for how solidly people in the workshop feel about the Python basics? If some of the Python concepts are difficult, moving into ideas about using Pandas is a very steep curve, but if it comes easy because it ends up populated with .e.g people with a lot of R experience, then that seems like a natural step. What if we go with gapminder and then gentle lead-in to using Pandas, focusing on Python concepts, as the lesson has us move, and start with a second-day syllabus about making programs readable / looking at more libraries: matplotlib, statsmodels, seaborn, numpy, and go from there depending on how we feel after Morning 1?

[bsmith89]

This sounds like a great syllabus. I'll come back and comment later today with a proposed list of topics for us to take directly from gapminder and/or customize from that base.

[bsmith89]

Morning 1:

• Same as the gapminder morning lesson through "plotting".
• Add 15-30 minutes during the first section, "Running and Quitting", to clarify how Jupyter relates to Python in the shell, how we start jupyter (i.e. cd to a directory, explain what our pwd is, use the jupyter command with the notebook argument)
• (Suggestion) Consider starting with a pre-built Jupyter notebook which summarizes everything they'll do that day: load data, slice it, plot it. This both motivates the lesson and gives you something for demonstrating how to executing code within the notebook environment. After the initial demo, though, you can open up a blank notebook and teach the lesson from scratch.

Morning 2:

• Lists/indexing (to contextualize/re-emphasize pandas indexing)
• Writing functions
• Variable scope
• Conditionals
• pandas advanced filtering/indexing/selection
• pandas group/apply/combine
• pandas missing data
• basic statsmodels (linear regression)
• seaborn demo (stripplot, boxplot)

What else/instead for day 2? Currently I'm skipping sections on loops and defensive programming. I'd like to include a little more "best practices" type stuff.

[aerenchyma]

This sounds great! Can come back with additional ideas re: morning2, though I think this might fill up a lot of time.

[bsmith89]

@aerenchyma Are we settled on the lesson plan?

My understanding of it: you'll teach all of day 1 from the SWC gapminder lesson. I'll take day 2 with a more customized lesson which demos pandas, matplotlib, statsmodels, etc., and which can be shaped by learners feedback on day 1.

[aerenchyma]

@bsmith89 Yes, sounds good!

[aerenchyma]

I'll try to arrive a little early so we can talk, but I'll aim to basically follow gapminder, and perhaps we can try to leave some time 12:15 - to discuss how the stuff from the day can move into the next morning?