Feedback

[andrewpbray]

Repo

• organization looks great
• work on your README so that it can serve now as a landing page for the work that you've done on this project: describe briefly the questions that you tackled, the data that you used, and then point the reader to your deliverables.

Poster

• Where did you get this template? It's a bit difficult to parse and the green gradient background is distracting.
• The largest font labels on this poster should be things like "Purpose" (or goals) and "Data" and "Results". Right now, it looks like a hodgepodge of plots - these need to be integrated into a narrative. You could include your questions inside the goals in an upper left panel, data/methods in the upper middle panel, and a summary of results in the upper right panel. The bottom panels could be dedicated to these plots that answer the question and feed into the results.
• You may want to credit Jeff Ramsay in the lower right corner.

White Paper

• You'll want to shrink some of these plots (including their font/label sizes) or combine them using gridExtra. Right now they tend to dominate the paper and make the narrative difficult to follow.
• In speculating why a particular structure exists in you plots, be careful. In this for example: "This could be an indication that planting efforts should be re-focused to include more large trees. However, it is possible that the city is at carrying capacity for large trees - only a limited number of streets, parks, and yards have room for them." I'd shy away from making recommendations since the folks at Urban Forestly will have a better sense of what could be driving these trends.
• With the index, it'll be worth thinking through the fact that in some scenarios (a small planting strip) a small tree is actually better. That is, it's difficult to formulate a single index for the whole city.
• Define your terms: what is low income? What is normal income? The phrase "normal" is also a bit loaded - middle?
• Have you tried making a scatterplot of average income on average tree index to get a spatial take on the effect that you're looking at temporally in the plantings and income data?
• If the above plot shows a trend, you could remake the pdx map to instead be map of residuals from the model that uses income to predict tree quality.
• Do you have a sense of how much overplotting you might be seeing in the maps? You might want to use crosses instead of fill-squares then add some transparency.
• In persistence, be sure to describe clearly how you measured it.
• For your persistence map, I would drop the income background. i'm still worried about the overplotting thing, so here's an alternative: instead of viewing all of the trees, used a 2D smoother to smooth the 0's and 1s on whether it persisted. This would result in a contour or heat map where we're looking at probability of persistence. Google around for terms like heat map, kernel smoother, and loess.