Notes from May/June 2021 workshop

[jorisvandenbossche]

Matplotlib:

• [x] Example about putting long code into a function for reuse -> add a docstring!
• [x] Add some actual small exercises on basic matplotlib and matplotlib+pandas!

Seaborn:

• [x] In one of the exercises (eg in bacterial case study), add to the exercise a step where you need to access one of the axes and do a customization with matplotlib (to repeat the link between seaborn and matplotlib in practice)

Visualization landscape:

• estimator was not yet used before, should we then use it here?
• Is seaborn also an implementation of grammar of graphics?
• Link to download the Oostende gps data is wrong (it links to acceleration data instead of gps)
• Link to "Bokeh plotting pitfalls datashader" page doesn't exist anymore
• Datashader example can maybe use hvplot instead of holoviews (https://hvplot.holoviz.org/user_guide/Geographic_Data.html#plotting-points)

Reshape:

• [x] title before excercises + re-read titanic data (in case they didn't run all cells above)

Air quality:

• [x] glob glob edit to pathlib

General idea: create case study without reshaping (eg for seaborn you don't need reshape if your data is already tidy), and leave reshape/tidy for third day, while on second day we can already practice groupby and seaborn

---

First day:

• pandas notebooks 1, 2, 3a and 4 (timeseries)
• start with bike count case study

Second day:

• Indexing (3a)
• Groupby
• Combining data
• Start with biodiversity processing case study
• (break)
• Visualization matplotlib
• Continue case study

[stijnvanhoey]

With respect to

Example about putting long code into a function for reuse -> add a docstring!

I dropped the example function as we do not really use it during course and it is not really the best example either.

[stijnvanhoey]

cfr.

estimator was not yet used before, should we then use it here?

I think this is fine, as the main focus here is demonstration purposes.

[stijnvanhoey]

closing this one, tackled as part of https://github.com/jorisvandenbossche/course-python-data/pull/17 or in new issue