To improve if there ever is a next time

[mauro3]

• explain & lay out the toy-reseach project at the beginning
• maybe consider making the model and the model-application to Breithorngletscher separate repos
• do Dependencies before Testing?
• say something about deleting projects with confidence (as they can be restored)
• make the hand-drawn flow-chart in the intro-talk better
• add one lecture on how the project could be packaged in docker and/or a vm
• distinguish between "exploratory phase" and "final submission" in L01 and L06 and maybe elsewhere
• vocabulary: only use one of main vs master-script

[mauro3]

One could think about providing more of the mass-balance model as is (like if one would use someone else's model). This would free some time and mental resources to work more on running the model with data. Git usage could still be practiced on the scripting part.

[mjacqu]

Geospatial processing:

• define xdem version in environment.yml
• 8:30 - 12:30 was barely enough time to get through the vector and raster data notebooks. For a less rushed experience, at the very minimum send full set-up and download instructions beforehand. Or do a full day instead.
• If time should allow working on "own" analysis/figure, a full day workshop would be needed.

[mauro3]

Link to https://library.ethz.ch/news-und-kurse/kurse/forschungsdatenmanagement-und-verwandte-themen.html ?
In particular https://library.ethz.ch/news-und-kurse/kurse/forschungsdatenmanagement-und-verwandte-themen/reproduzierbares-wissenschaftliches-rechnen-und-datenanalyse.html

[mauro3]

Link to https://py.geocompx.org/ ?