R package with advanced diagnostics to evaluate a Stock Synthesis model. Diagnostics include residual analyses, hindcasting and cross-validation techniques, and retrospective analyses.
I met with Phoebe Woodworth-Jefcoats last week to go over the figures in ss3diags and she had some suggestions as far as default settings to improve overall readability, clarity, and 508 compliance. Her suggestions are listed below:
Use Arial as the default font (Arial is the font for tech memos at PIFSC because it can type Hawaiian language properly)
Font size should be at least 10pt
For retrospective plots, consider using a color scale instead of different colors
Fully spell out "Ref" line to make it more clear in the legend
Indicate grey uncertainty shading in the legend
In hindcast plots, change white ("actual data") line to black and make dashed lines dashed with color
Spell out the x-axis label for likelihood plots
Use the color contrast checker to see if colors would pass for 508 compliance (right now, most fail)
Suggested color palettes include nmfspalette, viridis, or dark2
@iantaylor-NOAA I wanted to share Phoebe's suggestions incase any are helpful for r4ss. One suggestion that was specific to r4ss was to print the actual parameter name that is being profiled over (x-axis label) instead of the SS3 name on the likelihood plots. But since you can easily specify the x-axis label with the profile.label arg, maybe as long as people do that, its not a major issue for you to address? (I was just showing her the default plot.)
I met with Phoebe Woodworth-Jefcoats last week to go over the figures in ss3diags and she had some suggestions as far as default settings to improve overall readability, clarity, and 508 compliance. Her suggestions are listed below:
nmfspalette,viridis, ordark2Notes from our meeting can be found here.
@iantaylor-NOAA I wanted to share Phoebe's suggestions incase any are helpful for r4ss. One suggestion that was specific to r4ss was to print the actual parameter name that is being profiled over (x-axis label) instead of the SS3 name on the likelihood plots. But since you can easily specify the x-axis label with the
profile.labelarg, maybe as long as people do that, its not a major issue for you to address? (I was just showing her the default plot.)