Inventory of Feedback on Version 0.1: 2022-02-08

SOLV-Code commented 2 years ago

High Priority to implement

[x] default to reporting 30,50,70 BM Prob decline (or have a default options button)
[x] How much control over the MCMC do we want to build in at this stage (e.g. menu options for num chains, burn-in etc?) If so, need to first build these arguments into the calcPercChangeMCMC() function of the package. For now those are hardwired. -> DONE
[x] change time window selection to 2 inputs:
- Avg Gen for the DU
- Num Gen to use for Prob Decl

[ ] build in option to select which columns to use? pkg needs 3 columns (DU, Year, Abd) but could set it up such that input files can have wharever columns, and then user chooses " col Z for Abd". That would make it easier to test different versions of index data, for example (do it in the app, rather than having to generate and load different data files). DU and Year columns would still be mandatory.
[ ] outputs: pdf report, word report, or individual table/fig saving? Eventually probably all of the above, but which one first?
[ ] Focus just on 1 year at a time, or build in retrospective analyses as well?
[ ] currently don't have the MCMC diagnostic plots in the app. may need to restructure the pkg outputs to select individual plots. OR jst option to get pdf? -> trying to avoid storing full MCMC samples and retrieving them every time a setting changes.
[ ] Put in a "refresh" button for the MCMC, so that can easily test the effect of just rerunning with diff random seed? right now need to change one of the settings to the same thing
[ ] show influence of prior on estimate (Bayes factor)

[ ] Option to click and use pre-loaded data set
[ ] Option to save your settings for next time: maybe have a button to export the current input object as an RDS and then a tab to load it next time? Or write to csv?

carrieholt commented 2 years ago

Some more suggestions:

[ ] In legend, suggest using label “MLE percent change” instead of Det: pchange, and Bayesian percent change. Also, suggest changing prob decl to Probability % change is less than -30% = , Probability % change is less than -50% = , etc., Or add input: 30%, 50%, or 70% to be shown
[x] For MCMC diagnostics: Store data for kernel density plot for posteriors, and rhat values
[x] Priority step: add priors as input (with MCMC settings, e.g., burnin, thinning, etc.) -> have done this with #6 merge
[x] Add a few other DUs to sample data (e.g. salmon)
[x] Suggest Changing title: “Explore Annual Estimates” to “Decline Estimates”
[x] Suggest changing “Scalar” to “Scale of abundances”
[x] Under Fit settings, suggest spelling out, Fit to loge(abundances) or raw abundances. MLE vs Bayesian estimation
[x] “Select an End Year” instead of “Select And End Year”
[x] The information in the legend could also be put in table form in the “Summary” tab
[x] General Settings: data processing? (may not be needed for now) -> have moved some UI pieces there, need to discuss
[x] Add note in readme/disclaimer that the current version of "MetricsCosewic" package should be installed. -> this is not an issue for the online version, only for running locally, also added it to the opening page of the app.

SOLV-Code commented 2 years ago

SOLV-Code commented 2 years ago

[x] general settings: multiple BM value inputs -> this one seems to be tricky -> just do 3 inputs for now
[x] new table: ProbDecl
[x] plotPattern setting revisions
[x] new plot: plotDistribution
[x] new plot: plotBoxes
[ ] addFit setting revisions
[ ] FIX any mention of ForecastR