Haromize estimation error output with RstoxBase

[edvinf]

Standard output in reports for RstoxFDA is:

• point estimate
• SD/SE
• High (interval limit)
• Low (interval limit)

StoX bootstrap report somewhat different stats of sampling distribution, including for instance CV. We should use the opportunity with 4.0 and break of backwards comp to harmonize.

[sindrevatnehol]

There is several data deliveries from catch estimates for the assessment required by the data submitters. Data deliveries from FDA, but not limited, need to include:

• Report of abundance by quarter and length
• Abundance by age, quarter and ICES area (Possible in the current implementation)
• Mean length by age, quarter and ICES area
• Mean weight by age, quarter and ICES area
• Total catch in tonnes by quarter and ICES rectangle

[edvinf]

Harmonizing FDA reports with StoX bootstrap would entail:

• exchange the interval parameter with percentages parameter that provide quantiles for reporting (not mandatory, and not limited to two)
• make format less rigid so that one can have varying number of columns depending on which quantiles are requested. The format already allows varying number of columns to account for domain variables, so we need to device a way where the two kinds of columns are distinguished for downstream use. Naming convention will do, but it would be preferrable if we could enforce that domain variables are not introduced that violate the naming convention.
• provide cv in addition to sd
• provide unit as a column

In the FDA workflows, reports are input to plots. So plots would have to implement some checking that expected columns are provided, and some logic to deduce the highest and lowest interval.

Interpretation would depend on the process leading up to the plots, but that is already in the cards. These are summary statistic for a sampling distribution, which is either estimated by bootstrap (StoX bootstrap), Bayesian posterior (ECA), or approximating normal (analytical).

Check with @arnejohannesholmin that this list of changes is complete, and that there are no other additions expected in upcoming versions that we should take account of.