changes before resubmission

[gottacatchenall]

a running list of changes before resubmission: please feel free to add any changes from coauthors to this list

---

things to change:

• [x] ~consider changes to figure one (see next post in this thread)~ not doing this
• [x] remove figure 4
• [x] remove references to fig4 in validation section

• [x] citations in introduction have disappeared again

things to rephrase:

• [x] the second sentence of the abstract
• [x] the last sentence of the Case Study section ("As in fig. 2, this highlights how much data sources can be consumed by models predicting species interactions.")
• [x] beginning of What is a predictive model section ("In order to build a predictive model under the Bayesian paradigm...")
• [x] roadmap figure intro paragraph "Here we present a conceptual roadmap (fig. 2) which" needs a comma, or which -> that

[gottacatchenall]

re figure one changes:

1. in the network illustration repo (PoisotLab/NetworkLearningIllustration@b4c26f3118dbd2e3136e1d7ea0046bdbad418840 ) i've added a file to do many partitions of test/training sets to plot a density of the distribution of roc-pr curves. the code is ugly but i think this could help address review one comment #128 and is generally a good practice to put on display in the paper. the resulting roc-pr curves are below.

The downsides are that we don't have a way to indicate a "best" threshold, as indicated by Youden's J on the current version of figure 1. Also it takes about an hour to run on a single core laptop, which isn't great for example code with a paper. It does look neat though

2. also the network plot (the right side of current figure 1) has empirical interactions in purple and imputed interactions in grey, which may not be clearly visible and should be tweaked to ensure the difference is apparent. perhaps splitting the network plot into two panels (empirical interactions in one and imputed interactions in the other)?

[TanyaS08]

Okay where I'm at with figures (mostly because I've actually lost track of all the discussions):

Figure 1 (Proof-of-Concept) -> do whatever you feel is best/clearer

Figure 4 (AUC) -> Drop it (although there is enough residual attachment on my side that if you guys have a strong desire to keep it I'm not going to have a breakdown)

[TanyaS08]

Also @gottacatchenall feel free to work on #144 and make the edits you've suggested here (not all of them are immediately clear to me....)