Before finding Seattle's API data source I'd at first pulled their permits issued data via their excel sheets. This was kind of hassle... See related gist here: https://gist.github.com/brshallo/7a14235134f8e10139f71c3369f8d50f

permits-issued.md

library(tidyverse)

urls <- tibble(month = month.name, month_num = 1:12) %>% 
  cross_join(tibble(year = 2019:2023)) %>% 
  arrange(year) %>% 
  mutate(year_month = make_date(year = year, month = month_num)) %>% 
  filter(year_month < floor_date(today(), "months")) %>% 
  mutate(urls = paste0("https://www.seattle.gov/documents/Departments/SDCI/Resources/Stats/", year, month, "Summary", ".xlsx"))

url_to_units <- function(url, sheet = NA){
  
  if(is.na(sheet)){
    data <- openxlsx::read.xlsx(url, startRow = 6)
  } else {
    data <- openxlsx::read.xlsx(url, sheet = sheet, startRow = 6)
  }
  
  data %>% 
    as_tibble() %>% 
    janitor::clean_names() %>% 
    # filter(decision_type == "Grand Total") %>% 
    filter(decision_type == "New", dept_of_commerce != "Single_Family/Duplex") %>%
    mutate(across(starts_with("units_"), ~ifelse(is.na(.x), 0, .x))) %>% 
    summarise(new_added = sum(units_added - units_removed)) %>% 
    pluck("new_added")
}

safe_url_to_units <- safely(url_to_units)

data <- urls %>% 
  mutate(new_units = map(urls, safe_url_to_units))

### Fix inconsistencies in sheets for a couple months
did_safely_error <- function(safely_named_list){
  
  map(safely_named_list, "error") %>% 
    map_lgl(is.null) %>% 
    {!.}
}

errors <- data %>% 
  filter(did_safely_error(new_units))

data_sep2019 <- errors %>% 
  filter(month_num == 9, year == 2019) %>% 
  mutate(new_units = url_to_units("https://www.seattle.gov/documents/Departments/SDCI/Resources/Stats/2019SeptSummary.xlsx"))

data_feb2020 <- errors %>% 
  filter(month_num == 2, year == 2020) %>% 
  mutate(new_units = url_to_units(urls, sheet = "February Summary"))
### 

data_clean <- data %>% 
  filter(!did_safely_error(new_units)) %>% 
  mutate(new_units = map_dbl(new_units, ~.x[[1]])) %>% 
  bind_rows(data_sep2019, data_feb2020) %>% 
  arrange(year, month_num)

data_clean %>% 
  mutate(year_quarter_label = tsibble::yearquarter(year_month),
         year_quarter = ceiling_date(year_month, "quarters")) %>% 
  group_by(year_quarter, year_quarter_label) %>% 
  summarise(new_units = sum(new_units)) %>% 
  ggplot(aes(x = year_quarter, y = new_units))+
  geom_line()+
  theme_bw()