sdamerdji / rezoner

rezoner

Run the app locally

Here is the minimal set of instructions to get the app running locally.

In RStudio run the following commands:

required_packages <- c("shiny", "dplyr", "sf", "shinyjs", "shinyBS", "mapboxer", "sortable", "shinyWidgets", "stringr", "viridis", "compiler", "RColorBrewer")
missing_packages <- setdiff(required_packages, installed.packages()[, "Package"])

if (length(missing_packages) > 0) {
  install.packages(missing_packages)
}

If you have trouble installing a package, try this: install.packages('foobar_package', type='source')

Then in RStudio, open app.R and click 'Run App' at the top right. The app should run!

You do not need to source main.R to run the webapp. Sourcing main.R is only needed to reproduce light_model.rds, sf_map.rds, and five_rezonings_nongeo.rds, but you get these for free by git cloning this repository.

Reproduce collateral for web app

More libraries are needed to run through the whole pipeline to re-create the pipeline that outputs light_model.rds, sf_map.rds, and five_rezonings_nongeo.rds

First you'll need to install git-lfs

In the command line, run the following:

git lfs pull

Then, on top of the packages installed to run app.R (see the above section), install these additional R packages:

required_packages <- c("readr", "tidyr", "modelr", "caret", "sfarrow", "tidytransit", "doParallel", "lwgeom")
missing_packages <- setdiff(required_packages, installed.packages()[, "Package"])

if (length(missing_packages) > 0) {
  install.packages(missing_packages)
}

Then, you'll source main.R. Depending on how many cores your computer has, this may take over an hour to run.

Make changes to the web app

Most of the logic in this web app occurs in update_df_() in app.R. If you want to change how the effect of rezonings are calculated, this is the function to change.

If you want to define a custom rezoning with greater resolution than the UI allows, look at the function yimbycity() for a pattern to borrow. In particular, you can zone at the parcel level by using the block/lot combination of parcels, which appear in the popup when you click the parcel on the map.

The web app can easily be debugged by placing the line "browser()" at a line in the code where you'd like to set a breakpoint.