fw_zillow_check.R

## R script to check TAD building age information against Zillow API

library(foreign)
library(xml2)
library(dplyr)

dbf <- "fw_footprints_mb.dbf"

dat <- read.dbf(dbf, as.is = TRUE)

set.seed(1983)


# Get sample of addresses from the buildings DBF
samp <- sample(dat$ADDRESS, 1000, replace = FALSE)
  
# You can get your own API key from Zillow
api_key <- "XXX"

# Format the address for the API call
formatted <- gsub(pattern = " ", replacement = "+", samp)

# Function to pull data from the Zillow API
get_year <- function(address) {
  
  api_1 <- paste0("http://www.zillow.com/webservice/GetDeepSearchResults.htm?zws-id=", 
                  api_key, "&address=")
  
  api_2 <- "&citystatezip=Fort+Worth%2C+TX"
  
  api_string <- paste0(api_1, address, api_2)
  
  api_call <- read_xml(api_string)
  
  find_year <- xml_find_all(api_call, ".//yearBuilt")
  
  yb <- xml_text(find_year)
  
  return(yb)
    
}

# Call the function over the vector of addresses
yrs <- sapply(formatted, function(x) get_year(x))

# Clean up
vec <- unlist(yrs)

df <- data.frame(address = names(vec), yearBuilt = vec, row.names = NULL)

df$address <- gsub("\\+", " ", df$address)

df$address <- gsub("\\d*$", "", df$address)

dup <- duplicated(df$address)

df2 <- df[!dup, ]


# Check for matches
check <- inner_join(dat, df2, by = c("ADDRESS" = "address"))

check_unique <- check[!duplicated(check$ADDRESS), ]

matches <- check_unique$year_label == check_unique$yearBuilt

# See where the match is TRUE
summary(matches)

# Check to see how many unmatched addresses were within 10 years
unmatched <- check_unique[!matches, ]

unmatched$diff <- as.numeric(as.character(unmatched$yearBuilt)) - unmatched$year_built

summary(unmatched$diff < 10)