Get violin auction prices from a few websites and plot

violin-prices.R

library(rvest)

url <- "https://www.maestronet.com/history/makers_list.cfm?ID=1"


links_on_page <- 
url %>%
  read_html() %>% 
  html_nodes(".text a") %>% 
  html_attr('href')

pages_from_links <- 
  str_glue('https://www.maestronet.com/history/{links_on_page}')

safely_read_html <- 
  safely(read_html)

get_pages <- 
  map(pages_from_links, ~.x %>%  safely_read_html())

get_pages_results <- 
map(get_pages, ~.x$result) %>%  
  compact

get_pages_results_tables <- 
map(get_pages_results, ~.x %>%  
    html_nodes("td td td td") %>% 
    html_text) 

get_pages_results_prices <- 
  map(get_pages_results_tables, ~ str_subset(.x, "\\$.")) %>% 
  unlist %>% 
  parse_number() %>% 
  enframe

# plot distribution of auction prices from www.maestronet.com
ggplot(get_pages_results_prices,
       aes(value)) +
  geom_histogram() +
  geom_vline(xintercept = median(get_pages_results_prices$value),
             colour = "red",
             size = 2) +
  annotate("text",
           x = median(get_pages_results_prices$value) + 17000,
           y = 250,
           label = str_glue("Median price is ${median(get_pages_results_prices$value)}"),
           size = 6) +
  scale_x_log10(labels = scales::dollar) +
  theme_minimal(base_size = 16) +
  xlab(str_glue('Auction prices of {nrow(get_pages_results_prices)} violins\n data from https://www.maestronet.com/'))


# how about at https://tarisio.com/auctions/auction-results/

url_r <- "https://tarisio.com/auctions/auction-results/"

tarisio_com_auctions <- 
url_r %>% 
  read_html() 

tarisio_com_auctions_prices <- 
tarisio_com_auctions %>% 
  html_table(fill = TRUE) %>% 
  .[[1]] %>% 
  janitor::clean_names() %>% 
  select(starts_with('price')) %>% 
  stack %>% 
  mutate(value = parse_number(values)) %>% 
  filter(!is.na(value))

# plot distribution of auction prices from www.maestronet.com
ggplot(tarisio_com_auctions_prices,
       aes(value)) +
  geom_histogram() +
  geom_vline(xintercept = median(tarisio_com_auctions_prices$value),
             colour = "red",
             size = 2) +
  annotate("text",
           x = median(tarisio_com_auctions_prices$value) + 10000,
           y = 25,
           label = str_glue("Median price is ${median(tarisio_com_auctions_prices$value)}"),
           size = 6) +
  scale_x_log10(labels = scales::dollar) +
  theme_minimal(base_size = 16) +
  xlab(str_glue('Auction prices of {nrow(tarisio_com_auctions_prices)} violins\n data from https://tarisio.com/auctions'))

# and what about Christies

the_page <- "https://www.christies.com/lotfinder/searchresults.aspx?pid=temporarysoldlotsbanner&searchtype=p&action=paging&entry=violin&pg=all&sid=3666efe5-47bd-4a4a-a5b9-6eb10ea5e008"
the_node <- ".accept-cookies-button , h3 , .price"

christies_auctions <- 
  the_page %>% 
  read_html()  %>% 
  html_nodes(the_node) 

christies_auctions_usd <- 
christies_auctions %>% 
  html_text() %>% 
  str_subset("Price") %>% 
  str_subset("USD")  %>% 
  parse_number() %>% 
  enframe()

# plot distribution of auction prices from Christies
ggplot(christies_auctions_usd,
       aes(value)) +
  geom_histogram() +
  geom_vline(xintercept = median(christies_auctions_usd$value),
             colour = "red",
             size = 2) +
  annotate("text",
           x = median(christies_auctions_usd$value) + 15000,
           y = 60,
           label = str_glue("Median price is ${median(christies_auctions_usd$value)}"),
           size = 6) +
  scale_x_log10(labels = scales::dollar) +
  theme_minimal(base_size = 16) +
  xlab(str_glue('Auction prices of {nrow(christies_auctions_usd)} violins\n data from https://www.christies.com'))

  
  
  

Includes some bows and other things: