R Notebook

Load libraries and R Data from assignment 1

library(tidyverse)

load("/home/rstudio/Mod 1.RData")
#yelp_all (restaurants)
#yelp_all2 (hiking)
#yelp_all_c (combined)

first flatten the data frame to remove nested columns in location

yelp_flat_c$coordinates%>% head()

Warning: Unknown or uninitialised column: `coordinates`.

NULL

concatenate the transaction column

yelp_concat_c <- yelp_flat_c %>% 
  mutate(transactions = transactions %>% 
           map_chr(., function(x) str_c(x, collapse=", ")))
head(yelp_concat_c)

function to flatten the category column

# Custom function that takes the data frame in "categories" column in Yelp data
# and returns a character vector
concate_list_c <- function(x){
  # x is a data frame with columns "alias" and "title" from Yelp$categories
  # returns a character vector containing category concatenated titles 
  titles <- x[["title"]] %>% str_c(collapse = ", ")
  return(titles)
}

yelp_flat2_c <- yelp_concat_c %>% 
  mutate(categories = categories %>% map_chr(concate_list))

yelp_flat2_c %>% print(width = 1000)

removing duplicate values

yelp_unique_c <- yelp_flat2_c %>% 
  distinct(id, .keep_all=T)

glue::glue("Before dropping duplicated rows, there were {nrow(yelp_flat2_c)} rows. After dropping them, there are {nrow(yelp_unique_c)} rows") %>% 
  print()

Before dropping duplicated rows, there were 2818 rows. After dropping them, there are 749 rows

Are there any NA values

yelp_unique_c %>% 
  map_dbl(., function(x) sum(is.na(x))) 

                      id                    alias 
                       0                        0 
                    name                image_url 
                       0                        0 
               is_closed                      url 
                       0                        0 
            review_count               categories 
                       0                        0 
                  rating             transactions 
                       0                        0 
                   price                    phone 
                     208                        0 
           display_phone                 distance 
                       0                        0 
    coordinates.latitude    coordinates.longitude 
                       0                        0 
       location.address1        location.address2 
                       5                      161 
       location.address3            location.city 
                     263                        0 
       location.zip_code         location.country 
                       0                        0 
          location.state location.display_address 
                       0                        0 

  # map_dbl is a variant of map() which outputs 
  # numeric vector rather than a list.

There are 5 NA values in $location.address1 - and the other location address columns which don’t matter either because there are no missing values from lat/long columns. NA values in $price may not matter either for this study.

Try removing NA values in $price

yelp_dropna2_c <- yelp_unique_c %>% 
  drop_na(price)

This was bad, all the ‘hiking’ sites had NA values in price and I lost part of my data. UNDO

Checking for data cases outside of the Chatham County geographic border

# census boundary
census <- read_sf("/home/rstudio/Mod 1/chatham.shp") %>%
  st_transform(4326)


# Converting yelp_unique_c into a sf object
yelp_sf_c <- yelp_unique_c %>% 
  st_as_sf(coords=c("coordinates.longitude", "coordinates.latitude"), crs = 4326)
  
# sf subsets
yelp_in_c <- yelp_sf_c[census %>% 
                     st_union(), ,op = st_intersects]

Compare tidy and messy data

glue::glue("nrow before: {nrow(yelp_all_c)} -> nrow after: {nrow(yelp_in_c)} \n
            ncol before: {ncol(yelp_all_c)} -> ncol after: {ncol(yelp_in_c)} \n") %>% 
  print()

nrow before: 2818 -> nrow after: 709 

ncol before: 16 -> ncol after: 23 

Visualize

tmap_mode("view")

tmap mode set to interactive viewing

tm_shape(yelp_in_c) + tm_dots(col = "price")

More visualizations

tmap_mode("view")

tmap mode set to interactive viewing

tm_shape(yelp_in_c) + tm_dots(col = "rating")

##The arrangement of food business in the gridded streets of Savannah contrast sharply with the more organic shorelines of coastal Chatham county’s hiking trails and the inland restaurants in suburban Savannah. While this could partially be seen before, after tidying the data, the pattern becomes clearer, emphasizing the relationship between urban form and businesses. Looking at the relationship between price point and rating seems to indicate that the density of businesses allows for more high star/ high price point restaurants. The three main pockets of high review count businesses are downtown Savannah, near the Savannah/ Hilton Head Airport, and on Tybee Island, a popular vacation destination. The largest grouping of 3 dollar sign restaurants occurs in downtown Savannah, especially by the waterfront, and in what is considered the most touristy area. One thing that is interesting is that looking at the map of Chatham county zoomed all the way out, you can only see 1 dollar sign and 2 dollar sign restaurants, while looking at the ratings map, there appear to be more 3-4 and 4-5 star restaurants than any other category. This indicates that Chatham county has a good value money in it’s dining scene. ##