Comments on 9. Visualizations

[TristanGarry]

Visualizations

Comments based on SW carpentry's Visualizing Tabular Data

• Import aliases (e.g. import matplotlib.pyplot as plt, import numpy as np) should probably be encouraged from the start as it's more common with these libraries and would make it easier for people taking the course to debug their own errors using Stackoverflow or other methods. Since none of the examples include aliases and they are mentioned at the end some students could end up not implementing it in their own code in the future.

• Would teaching seaborn alongside matplotlib be in scope? It's often more user-friendly to use seaborn to generate initial plots and then access the matplotlib API to fine-tune plots, and depending on the application many users won't ever have to go past the seaborn API.

• As discussed in the last meeting, moving functions into this unit would make sense

  • An intro function could be some simple function for determining a Y limit with some padding (e.g. + 5%)
  • Then lead into putting the plot grouping part into a function

Other

• Is floating-point approximation in-scope for this course? Some people might be coming from software that handles decimals elegantly for them and might not even think to check that their decimal values are accurate in Python, this is especially important in the context of official statistics. A fun example for this problem is: Using Python, prove that 0.1 + 0.2 = 0.3

[alexmcsw]

@TristanGarry Thanks for the comments! I have a couple quick notes here but if you could start to incorporate these changes into a pull request that would be great :)

1. I agree about the aliases, definitely use them here and we can add a quick intro to the 2nd lesson?
2. I don't know if we want to teach multiple methods-- it can become very confusing quickly. I would suggest to keep at matplotlib for now and maybe at the next meeting we can get a pulse check on adding in sns?
3. sounds good for the functions
4. That's a good idea about floating point. Would that be more appropriate in one of the earlier lessons about variable types?