[SFAII] improving reports with additional graphics

[awig]

• add reliability diagrams for probabilistic forecasts
  • be careful night values can really skew the results
• histogram of errors
• other ideas?

[awig]

There are functions in plotly_figures.py that were never implemented. Documenting my thoughts on them and happy to get more input.

The functions are:

• nested_bar : can't picture what you would nest, maybe for combining categories and totals? or maybe multiple bar charts in single figure (e.g. showing all season forecasts 2x2)?
• joint_distribution : maybe this was intended just for computation, but is there are plot that we might want?
• marginal_distribution : same question as joint distribution
• taylor_diagram : i believe it would be used to show between one or more std of forecasts (y-axis) and the std of observations (x-axis) and then the correlation coefficient (radial-axis)
• probabilistic_timeseries : not sure what the thought specifically was here, suppose it would be similar to timeseries function which just scatter plot. maybe just an alternative to show nicer timeseries plots if they are timeseries (say with opacity and fill around the 50% interval)
• reliability_diagram : will be added
• rank_histogram : think this would be relative frequency (y-axis) and bins of error intervals (x-axis). would errors be normalized or not, how to deal with outlier errors when that skews the binning?

[awig]

Example: