5 Using comments to trigger inline code

Code completion is predicting your next line from your script (and any other open scripts) — it won’t know your variable names unless you’ve provided them somewhere. One way is to keep a readme.md open with your metadata in it. Another, which tends to work more reliably, is to write a comment describing what you want directly above where you’re typing:

# Make a point plot of secchi against pres.topa
gg...

That should get you the right ggplot. Using the actual variable names in your comment is more precise and helps the assistant guess correctly — this is the same “be specific” principle that shows up everywhere in this book, just applied one line at a time.

You can also put key variable names in a comment block at the top of your script, so every completion after that has them in view.

Another way to use this is to not write any R at all — just write a series of comments and let the assistant fill in the code. Try templating a series of plots like this:

# Make a point plot of secchi against pres.topa with a stat_smooth

# Plot logged (two categories) and pres.topa as a boxplot

# Plot CB_cover (branching coral cover) against secchi

Remember one way to improve prompts is to be more specific. Using the correct variable names helps, whether you’re prompting a chat window or just a code comment.

Now go back through and click under each comment to trigger the suggestions.

This strategy is especially handy in data wrangling. As a simple example, try producing this grouped summary using comments only:

dat %>%
    group_by(logged) %>%
    summarize(mean_topa = mean(pres.topa),
                mean_CB = mean(CB_cover))

To get there I’d write something like:

    # group dat by logged
    # summarize pres.topa and CB_cover

If you’ve already documented the variable names above (in a readme or a comment block), you can often be lazier and less precise here and still get a correct completion.

Challenge

Using comments only (no R code typed by you), get the assistant to produce a summary table of mean secchi and mean pres.topa grouped by flow. Then try the same task with the variable names left out of your comments — compare how many attempts it takes.