Define analysis/plots required for the application notebook

[fnattino]

Send Raul links to available notebooks so that he can point out which analysis are still missing

[fnattino]

Collected the main notebooks which we have run in the past to perform the co-clustering analysis for the CONUS dataset (or a subset of it) and asked for feedback to Raul and Emma:

• CGC tutorial (this is the one that includes most details and description of the steps)
• Example notebook in the CGC repository (the CGC tutorial above has been actually based on this notebook)
• Notebook that has been run on the AWS infrastructure (this notebook does not have much post-processing plots, but it includes some visualization of the input dataset based on Xarray)

[fnattino]

Got feedback from Raul:

Feedback notebook eScience Academy. Figures [18] and [19]. Axis and color bar (legend) are unclear. If the data is reordered, how come both the x and y axis grow monotonically? The legend reaches values that are too high (225) and, as a result, the co-clusters are not clearly visible. Max and min of the legend should be similar to the ones used in figure [22]. K-means features need to be exposed. Selection of optimum K? (I wonder if a threshold is the best approach to identify the elbow vs change in slope?) Temporal patterns are not visible yet. Figures 8 and 10 from this publication No need to immediately implement those figures. Need to think about the case study.

[fnattino]

Fixed all issue apart from the discussion on the selection of optimum K in k-means, whiche requires also input from Emma.