apple-app-store-rankings-draft.R

# import libraries
library(aws.s3)
library(tidyverse)
library(jsonlite)
library(stringr)
library(ggbump)
library(cowplot)

# setting keys
Sys.setenv("AWS_ACCESS_KEY_ID" = "AWS_ACCESS_KEY_ID", 
           "AWS_SECRET_ACCESS_KEY" = "AWS_SECRET_ACCESS_KEY",
           "AWS_DEFAULT_REGION" = "us-east-1")

# get all files within specific directory of bucket
all_files = get_bucket('egw-data-dumps', prefix = 'apple-app-store')

# loop over to extract file names
paths = c()
for (i in 2:length(all_files)) { paths = c(paths, all_objects[i]$Contents$Key) }
paths = paste0('s3://egw-data-dumps/', paths)

# get objects, compile dataframe
datalist = list()
for (i in 1:length(paths)) {
  print(paths[i])
  obj = get_object(paths[i])
  datalist[[i]] = read_csv(rawToChar(obj)) %>% mutate(path = paths[i])
}
raw = do.call(rbind, datalist)

# clean things up a bit
chart_history = raw %>% 
  mutate(app_id = basename(link)) %>%
  mutate(rank = as.numeric(str_extract(label, '[0-9]+'))) %>%
  mutate(developer = sub("\\.$", "", str_remove(str_remove(label, '([^.]+?)\\. '), '([^.]+?)\\. '))) %>%
  mutate(app = sub("\\.$", "", trimws(str_remove(str_remove(label, developer), '([^.]+?)\\. ')))) %>% 
  select(rank, app, developer, link, scraped_at, label, path)

# check thigns out
chart_history %>% filter(rank == 26) %>% View()
chart_history %>% group_by(label) %>% count(sort = T) %>% View()
chart_history %>% filter(developer == 'Crypto') %>% View()

# https://github.com/davidsjoberg/ggbump
rank_threshold = 10
chart_history %>%
  #filter(rank <= rank_threshold) %>%
  #filter(developer == 'Daily Pay Inc.') %>%
  #filter(developer %in% c('Binance LTD', 'Coinbase Wallet', 'Crypto.Crypto.com')) %>%
  filter(developer %in% c('Albert Corporation', 'Daily Pay Inc.')) %>%
  ggplot(., aes(x = scraped_at, y = rank, color = app)) +
  geom_point(size = 2) +
  #geom_text(data = chart_history %>% filter(scraped_at == min(scraped_at) & rank <= rank_threshold), aes(x = scraped_at - .1, label = developer), size = 3, hjust = 1) +
  #geom_text(data = chart_history %>% filter(scraped_at == max(scraped_at) & rank <= rank_threshold), aes(x = scraped_at + .1, label = developer), size = 3, hjust = 0) +
  geom_bump(size = 1, smooth = 15) +
  theme_minimal_grid(font_size = 14, line_size = 0) +
  theme(legend.position = "none", 
        panel.grid.major = element_blank()) +
  scale_y_reverse()

chart_history