Gist com análise do tiktok

tiktok.R

library(tidyverse)
library(stringr)

# CRIA LISTA DE CSVS
# full.names = F para tirar nome do diretorio
csvs <- list.files("20210811", pattern = "*.csv", full.names = FALSE)

setwd("20210811")

# CRIA TABELAO COM COLUNA DE NOME DE ARQUIVOS
tabelao = tibble(File = csvs) %>%
  extract(File, "hash", remove = FALSE) %>%
  mutate(Data = lapply(File, read_csv)) %>%
  unnest(Data) %>%
  select(-File)

setwd("../")

# CONSERTA DATAS DO FORMATO UNIX PRA ALGO MAIS HUMANO
tabelao$data <- as.Date(as.POSIXct(tabelao$createTime, origin="1970-01-01"))


# diggCount = likes
df <- tabelao %>%
  select(hash, id, data, text, authorMeta.name, authorMeta.verified, authorMeta.following, authorMeta.fans, webVideoUrl, musicMeta.musicAuthor, musicMeta.musicOriginal, videoMeta.duration, diggCount, shareCount, playCount, mentions, hashtags) 
  # distinct(id, ..., .keep_all = FALSE)

videos <- df %>% 
  group_by(hash) %>%
  count() %>%
  arrange(desc(n))

videos_uniques <- df %>% 
  group_by(id) %>%
  count() %>%
  arrange(desc(n))

plays <- df %>% 
  group_by(hash) %>%
  summarise(median_plays = median(playCount)) %>%
  arrange(desc(median_plays))

likes <- df %>% 
  group_by(hash) %>%
  summarise(median_likes = median(diggCount)) %>%
  arrange(desc(median_likes))

concat <- left_join(likes, plays, by = "hash") %>% left_join(., videos, by = "hash")

library(clipr)
write_clip(concat)

mentions <- df %>% 
  group_by(hash) %>%
  count(mentions) %>%
  arrange(desc(n))

stats_base <- df %>%
  summarise(data_min = min(data), 
            data_max = max(data), 
            plays_min = min(playCount),
            plays_max = max(playCount),
            likes_min = min(diggCount),
            likes_max = max(diggCount)
            )

# concatena dados com categorias manuais em df separado
concat_categorias <- left_join(df, d, by = "hash")

militarismo <- k %>%
  select(text, webVideoUrl, categoria) %>%
  filter(categoria == "policial/militar")

# POPULARIDADE

k <- concat_categorias

#d$data_publ <- lubridate::as_date(d$`Post Created Date`, format="%Y-%m-%d")

contagem <- k %>%
  filter(categoria != "memes") %>%
  filter(categoria != "esporte") %>%
  filter(data >= "2020-01-01") %>%
  mutate(ano_mes = format(data, "%Y-%m")) %>%
  group_by(categoria,ano_mes) %>%
  count(data)

contagem$rede <- "TikTok"

contagem %>% 
  drop_na() %>%
  ggplot(aes(ano_mes, n, fill = categoria, colour = categoria)) + geom_bar(stat = "identity", position = "dodge")

# mediana
mediana_plays <- k %>%
  filter(categoria != "memes") %>%
  filter(categoria != "esporte") %>%
  filter(data >= "2020-01-01") %>%
  mutate(ano_mes = format(data, "%Y-%m")) %>%
  group_by(categoria,ano_mes) %>%
  summarise(mediana = median(playCount))

mediana_plays %>% 
  drop_na() %>%
  ggplot(aes(ano_mes, mediana, fill = categoria, colour = categoria)) + geom_bar(stat = "identity", position = "dodge")