walkerke / census-with-r-book Goto Github PK

Tracking typos here as I find them. Readers: feel free to let me know here!

• No chapter reference link to Ch9; incorrect link to Ch10 https://walker-data.com/census-r/analyzing-census-microdata.html#pums-data-and-the-tidyverse
• Should say the 2010 Brazilian Census, not 2020 https://walker-data.com/census-r/analyzing-census-microdata.html#pums-data-and-the-tidyverse
• Broken reference due to misspelling https://walker-data.com/census-r/analyzing-census-microdata.html#pums-data-and-the-tidyverse

Hello

I am working with Ch. 7 in your book and I wanted to run the distance and proximity analysis you have in chapter 7.4. Specifically travel time.

I am bringing and csv file (destinations) and US Census block groups (centroids). When I run the mb_matrix function I get the following error:

Attached is the csv file for destinations.

I wonder if in the problem is in the geocoding process.

FamilyChildCareAndCentersAndPreschools.csv

I used tidygeocoder to geocode the destinations

childcare <- read.csv("D:/Googledrive/COE/FamilyChildCareAndCentersAndPreschools.csv") 

childcare$fulladdr <- paste(as.character(childcare$Street),
                            as.character(childcare$City),
                            as.character(childcare$State),
                            as.character(childcare$Zipcode))

geoCodechildcare <- geocode(childcare, address = 'fulladdr',
                            lat = latitude, long = longitude, method = "arcgis")

childcareSF <- st_as_sf(geoCodechildcare,
                        coords = c("longitude", "latitude"), na.fail = TRUE,
                        crs = 4326)

childcareSF <- st_transform(childcareSF, crs = 6501)


stearns_distances <- block_groups("MN", "Stearns", cb = TRUE)

st_crs(stearns_distances)

CRS.new <- st_crs("EPSG:6501")

stearns_6501 <- st_transform(stearns_distances, CRS.new)

library(mapboxapi)

# mb_access_token("pk............, install = TRUE)

# readRenviron("~/.Renviron")

times <- mb_matrix(stearns_6501, childcareSF)`

Running code in text localmoran _perm only generates 5 Column tibble data and next step in pipeline generates error. Have to manually reduce column names in 7.7.3 as no pi_simfolded,skewness, kurtosis columns. ? Hidden parameter

Hi Kyle,
here's a small suggestion for 8.1.

Instead of

ca_urban_data %>%
  split(~urban_name) %>%
  imap_dfr(~{
    .x %>%
      filter(variable %in% c("white", "hispanic")) %>%
      dissimilarity(
        group = "variable",
        unit = "GEOID",
        weight = "estimate"
      )
  }, .id = "urban_name") %>%
  arrange(desc(est))

I'd propose

ca_urban_data %>%
  filter(variable %in% c("white", "hispanic")) %>%
  group_by(urban_name) %>%
  group_modify(~
      dissimilarity(.x,
        group = "variable",
        unit = "GEOID",
        weight = "estimate"
      )) %>% 
  arrange(desc(est))

This doesn't require mixing Base R (split) and tidyverse, and I find it more naturally that the operation is done on the grouped data frame. It's really only stylistic though, so feel free to close.

4.5.2 Designing and styling the population pyramid

When I run the last code block in 4.5.2 I get:

Error: Breaks and labels are different lengths
Run rlang::last_error() to see where the error occurred.
In addition: Warning message:
Removed 30 rows containing missing values (position_stack).

FYI, the errors go away when I #comment out the following lines:

labels = ~ number_format(scale = .001, suffix = "k")(abs(.x)),

scale_y_discrete(labels = ~ str_remove_all(.x, "Age\\s|\\syears")) + 

This looks great, so happy to see it all put together!

At lunch with @dtburk, we talked about how slow it sounded to import the 1910 full count data in section 11.2.2.

I was curious and so figured I'd report back, but please don't feel like you need to add another caveat about workflow. Also, I still remain mystified by database administration, so there may be drawbacks I'm unaware of. Plus dBeaver is a nice thing to show to help people get comfortable with databases.

Those caveats aside, in my experimentation, I found chunked reading to be a lot faster than dBeaver's import. On my machine it was probably closer to 2.5 hours through dBeaver, but this takes ~15 minutes, loading only 10k rows at a time (it also works with both csv and fixed width extracts).

read_ipums_micro_chunked(
  "usa_00004.xml", 
  IpumsSideEffectCallback$new(function(x, pos) {
    dbWriteTable(
      conn, Id(schema = "ipums", table = "census1910"), x, append = TRUE
    )
  })
)

• In Chapter 7, Mapbox is no longer allowing depart_at for the driving profile; I'll need to change to driving-traffic.
• In Chapter 5, I should modify to show that 2021 shapefiles are available

Rebuild the book and add more as needed