In Metropolitan-Council/economic-values-atlas: Economic Values Atlas for the Twin Cities

knitr::opts_chunk$set(echo = FALSE, message = F, warning = F)

library(readxl)
library(tidyverse)
library(sf)

source("./data-raw/user_tracts_fxn.R")

Overview

The Economic Values Atlas project arises from work done by Oregon Metro and Brookings Institution.

MetCouncil's contributions

We have tried to make an RShiny application that is highly portable, well documented, and requires minimal coding in order to lower the bar for other regions who might like to implement this type of analysis.

At the most basic level, users may upload an Excel document containing 2 sheets. Users can also leverage R scripts to aggregate disparate data sources.

To understand what you will be creating, please view this example for the Twin Cities region or this example for the Portland region.

Set user parameters

First, set the following parameters which indicate the state/s and county/ies from which tract data comes from. Right now, this code is set to handle up to 2 states although this can easily be expanded if there is a need. Also set the metro name, and the format of the data inputs. This should be the only section of code that needs editing.

state_1 <- "OR"
county_1 <- c("Clackamas", "Columbia", "Multnomah", "Washington", "Yamhill")

state_2 <- "WA"
county_2 <- c("Clark", "Skamania")

metro <- "pdx"
dataformat <- "excel"

#####

# state_1 <- "MN"
# county_1 <- c("Anoka", "Carver", "Dakota", "Hennepin", "Ramsey", "Scott", "Washington")
# 
# state_2 <- NA
# county_2 <- NA
# 
# metro <- "msp"
# dataformat <- "rscript"

Read and process raw data

If the data is in an excel format, please ensure it has the following structure:

• Sheet 1: this sheet should be a "variable key" and contain the following columns:
• variable: a short code corresponding to the tract-level data used in the EVA
• name: a descriptive name corresponding to the tract-level data
• type: indicating if the variable corresponds to people, place, or business
• interpret_high_value: use high_opportunity if a high value of the variable should correspond to a positive economic value. use low_opporunity if a high value of the variable is not a desirable economic value.
• Sheet 2: this sheet should contain the raw data
• tract_string: should be the tract identifiers
• all other columns should be named according to the variables in sheet 1.

The excel file should be placed in the data-raw folder. It should be named according the the following convention: metro.xlsx. (Notice that Portland Metro's data had a different format/structure, so I processed that data differently; the suggested format here should hopefully save you from Excel headaches!)

If an R script is being used to aggregate the data, you may find it useful to follow the example for the Twin Cities outlined in the input_tract_data.R script.

#the following code processes the data. No edits are needed for end users.

### dl geometries using tigris
eva_tract_geometry <- user_tracts_fxn(state_1, county_1,
                               state_2, county_2)  

usethis::use_data(eva_tract_geometry, overwrite = TRUE)

map_centroid <- eva_tract_geometry %>%
  st_union() %>%
  st_centroid()
usethis::use_data(map_centroid, overwrite = TRUE)


###process tract data
if(metro == "pdx") {
  source("./data-raw/user_process_pdx.R")
} else if (dataformat == "excel") {
  eva_vars <- read_xlsx(paste0("./data-raw/", metro, ".xlsx"),
                       sheet = 1)

  eva_data_main <- read_xlsx(paste0("./data-raw/", metro, ".xlsx"),
                              sheet = 1,
                             na = c("NA")) %>% 
    mutate(tract_string = as.character(tract_string)) %>% #ensure this reads as a character so that we can join with tract geometries 
    gather("variable", "raw_value", -tract_string) %>%
    group_by(variable) %>%
     mutate(MEAN = mean(raw_value, na.rm = T),
         SD = sd(raw_value, na.rm = T),
         MIN = min(raw_value, na.rm = T),
         MAX = max(raw_value, na.rm = T),
         COUNT = as.numeric(sum(!is.na(raw_value))),
         z_score = (raw_value - MEAN)/SD) %>%
    right_join(eva_vars) %>%

    #create nominal weights
  mutate(weights_nominal = case_when(interpret_high_value == "high_opportunity" ~ (raw_value - MIN) / (MAX - MIN) * 10,
                                     interpret_high_value == "low_opportunity" ~ 10 - (raw_value - MIN) / (MAX - MIN) * 10,
                                     TRUE ~ NA_real_)) %>%

  #Weights Standard Score
  mutate(weights_scaled = case_when(interpret_high_value == "high_opportunity" ~ pnorm(z_score) * 10,
                                    interpret_high_value == "low_opportunity" ~ (10 - pnorm(z_score) * 10),
                                    TRUE ~ NA_real_)) %>%

  #weights rank
  mutate(weights_rank = case_when(interpret_high_value == "high_opportunity" ~ min_rank((weights_nominal)) / COUNT * 10,
                                  interpret_high_value == "low_opportunity" ~ min_rank(desc(weights_nominal)) / COUNT * 10,
                                  TRUE ~ NA_real_)) %>%

  # #rank
  mutate(overall_rank = case_when(interpret_high_value == "high_opportunity" ~ min_rank(as.numeric(weights_nominal)),
                                  interpret_high_value == "low_opportunity" ~ min_rank(desc(as.numeric(weights_nominal))))) %>%

  #clean dataframe
  select(-MEAN, -SD, -MIN, -MAX)

} else if (dataformat == "rscript") {
  source("./data-raw/input_tract_data.R")
}

usethis::use_data(eva_data_main, overwrite = TRUE)

Edit and add any region-specific langauge

There will likely be region-specific information that should be displayed alongside the data within the interactive application. A general shell is created here. But users should edit the ./R/mod_home.R script for any introductory information which should be displayed. And the ./R/app_ui/R script can be edited as well.

Text and pictures may also be updated, and the interface can be styled with css. I'm not sure of the best way to make that portable. My inclination is to create a "Brookings" style for the generic app.

Launch the app

The following code will launch your region's app (!!). Please run all the code chunks (two) prior to this section in order to see proper performance. To deploy it on an R server, you can click the blue button on the top right hand side of the app that will launch locally.

pkgload::load_all(export_all = FALSE, helpers = FALSE, attach_testthat = FALSE)
options("golem.app.prod" = TRUE)
eva.app::run_app() # add parameters here (if any)

Future plans

What are our future plans?!?!

• How can we update the app (particularly the data inputs) to suit our region?
• Are there national sources of supplemental data which might be nice to include (transit for instance?). If not, is this a thing that some code would still be helpful?
• Is z-score the best variable here? What are the other tabs in the Portland data?

Metropolitan-Council/economic-values-atlas documentation built on Dec. 17, 2021, 4:13 a.m.