New class plotter: best way to code

[AndreaGarciaJuan]

I used two new input variables in the new class Plotter:

• data_type (in init function): to know if data is gridded or individual profiles and change the type of plot

• co (in spatial_distribution an plot_posteriors functions): a dictionary to know the exacte name of latitude and longitude coordinate names in the dataset

Do you think this is the best way to code it? How do you do in pyxpcm to determine the name of latitude and longitude coordinates in dataset?

[gmaze]

* data_type (in _init_ function): to know if data is gridded or individual profiles and change the type of plot

can't be this inferred by the class ?

The dataset will be either [Z, N_PROF] for a collection of profiles or any combination of X,Y,T for the others (and possibly also including ensemble dimensions). So, basically, once we remove the vertical dimension, if we're left with 1 other dim, we have a collection of profiles, and if we have something else, we have a "gridded" product.

[gmaze]

* co (in spatial_distribution an plot_posteriors functions): a dictionary to know the exacte name of latitude and longitude coordinate names in the dataset

this should also come as optional: the Plotter class should look for CF standard axis 'X' and 'Y' internally first. If not found, then an error is thrown to request for explicit latitude, longitude coordinate names

they are many grids possible, so we should be precise to document here which kinds are supported.

[gmaze]

another question is about the integration with the existing m.plot class.

Should this new helper come as a separate module, like from pyxpcm import Plotter ?

Or should it come as a parent class leading to for eg m.plot.spatial_distribution() ?

[AndreaGarciaJuan]

I agree with your suggestions. I created a branch dev-plotter to code the modifications. I will also use the branch to develop the application of temporal representation function to 3D gridded data (not working at the moment). And to add a bar plot by year. if you have more suggestions feel free to share them with me

[AndreaGarciaJuan]

List of developments to be donne in class Plotter

Priority

• [x] Complete functions doc strings and definition of inputs
• [x] Automatic detection of coordinates
• [x] temporal_distribution function with gridded data

Non priority

• [x] for 3D gridded data including time create a function to get the most frequent label and then apply temporal_distribution function
• [x] plot together in the same figure all median profiles of each class