Visualiser tracking issue

[erolm-a]

With the end of this milestone comes the second alpha release of this project. This is a good moment to reflect on what we need to do with the generated data.

What we would like is a visualiser to show past run data, and possibly some tool to make a proper report out of that data. Ideally a PDF output or an ODS spreadsheet (for biologist stakeholders) would be perfect.

For now the wished report data should be:

• Gross fish/lice population during the time
• Breakdown into different genotype groups
• Event analysis, e.g. when treatment is applied, cage movements
• (Possibly) a nautical map to display lice movement across farms.

It would be nice to have a fully fledged GUI with controllable lines and elements rather than a simple static plot. For this reason, I'd suggest using Qt as its charting skills are very nice and customisable. This also puts us in the right mindset when trying to scale up the project and make the GUI more interactive.

Issues:

• For now, logging is purely done via JSON serialisation. This is good for humans, but it would require making a separate json decoder and rely a lot on unsafe dictionaries/rely on class getter tricks. It would be better to use pickle for serialisation purposes as it is by one order of magnitude faster than json and makes (de)serialisation a breeze. We can still retain JSON for human output while debugging - possibly by adding some verbose flags?
• As of now, the CLI needs to be revamped a bit. Automatically generate override flags for each property in params/test.json #71 would be nice to have at the same time
• More probes onto the queues may need to be added.

EDIT: here is a TODO list that summarises what was said above:

• Main visualiser UI
• Single simulation output
  • Farm population
  • Farm genotype distribution
  • Cumulative payoff output
  • Fish output per farm
  • Cage-specific output
  • Cage-averaged output
  • Convolution
  • Light mode
• Optimiser output
  • Probability space display
  • Payoff output
• Offspring
  • New offspring
  • Farm-to-farm movements
• Miscellanea
  • Configuration
    • Tree-based configuration
  • Embedded qtconsole
• Retractable panes/fixed aspect ratio

[erolm-a]

@magicicada @anthonyohare quick design questions: the plots currently look like this:

(in case you were wondering I only made a log plot of the lice population and not the rest)

My questions are:

• is the aspect ratio alright to you?
• what about the stroke styles and the colours? are they accessible enough? should I make the lines thicker?
• is a log plot appropriate in this case? Honestly I couldn't see anybody using log plots in the papers we have reviewed so far but nobody showcased gross lice populations so the need simply never arose?
• biology papers usually use dates/week counters rather than day counters; should I change the ticks this way?
• what else would you add/remove/change?

[magicicada]

Aspect ratio: seems fine to me on the individual plots, though we may want to reorganise the plots in a paper.
Stroke styles: thicker lines if possible. Aviod red/green if you can - maybe red/blue/orange?
log plot: use only if it makes things much more visible in a way that is important. Here might be worth it on the left, if we think that the differences early are important.
Dates only if we use seasonal weather or the absolute dates are meaningful for some other reason, otherwise I enjoy days as you have (but with added axis label)

Other things: axis labels, and ideally larger font for labels.

[erolm-a]

@magicicada what about this?