Navigation Menu

Skip to content

Instantly share code, notes, and snippets.

@rodolfoalmeida01

rodolfoalmeida01/tiktokers_topo.R

Created March 29, 2022 22:05
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save rodolfoalmeida01/b9b18b0fa817bc72e832136847c1b883 to your computer and use it in GitHub Desktop.
Save rodolfoalmeida01/b9b18b0fa817bc72e832136847c1b883 to your computer and use it in GitHub Desktop.
Download ZIP
Código para classificar contas de TikTok e gerar gráficos
Raw
tiktokers_topo.R
library(tidyverse)
library(plotly)
library(ggrepel)
# LENDO DADOS ----------------------------------------------------------------
# DISPONÍVEIS EM https://www.kaggle.com/moraesvic/tiktok-data-from-most-followed-accounts-2022
tt_data <- read_csv("~/tt_data.csv")
# CLASSIFICANDO -------------------------------------------------------
# Classifica categoria das contas
tt_data <- tt_data %>%
mutate(categoria = case_when(grepl("@happydogs2021", account_name, ignore.case = F) ~ "Animais",
grepl("@avani", account_name, ignore.case = F) ~ "Influencer",
grepl("@tiktok", account_name, ignore.case = F) ~ "Institucional",
grepl("@anokhinalz", account_name, ignore.case = F) ~ "Influencer",
grepl("@babyariel", account_name, ignore.case = F) ~ "Influencer",
grepl("@bellapoarch", account_name, ignore.case = F) ~ "Artista",
grepl("@brentrivera", account_name, ignore.case = F) ~ "Youtuber",
grepl("@charlidamelio", account_name, ignore.case = F) ~ "Influencer",
grepl("@briandadeyanara", account_name, ignore.case = F) ~ "Influencer",
grepl("@bts_official_bighit", account_name, ignore.case = F) ~ "Artista",
grepl("@carlosferiag", account_name, ignore.case = F) ~ "Youtuber",
grepl("@cznburak", account_name, ignore.case = F) ~ "Comida",
grepl("@darianrojasc", account_name, ignore.case = F) ~ "Influencer",
grepl("@dixiedamelio", account_name, ignore.case = F) ~ "Influencer",
grepl("@dobretwins", account_name, ignore.case = F) ~ "Influencer",
grepl("@domelipa", account_name, ignore.case = F) ~ "Influencer",
grepl("@elrodcontreras", account_name, ignore.case = F) ~ "Influencer",
grepl("@gilmhercroes", account_name, ignore.case = F) ~ "Comédia",
grepl("@homm9k", account_name, ignore.case = F) ~ "Comédia",
grepl("@gordonramsayofficial", account_name, ignore.case = F) ~ "Comida",
grepl("@itsjojosiwa", account_name, ignore.case = F) ~ "Artista",
grepl("@jamescharles", account_name, ignore.case = F) ~ "Maquiagem",
grepl("@junya1gou", account_name, ignore.case = F) ~ "Comédia",
grepl("@jasonderulo", account_name, ignore.case = F) ~ "Artista",
grepl("@joealbanese", account_name, ignore.case = F) ~ "Influencer",
grepl("@khaby.lame", account_name, ignore.case = F) ~ "Comédia",
grepl("@addisonre", account_name, ignore.case = F) ~ "Influencer",
grepl("@justmaiko", account_name, ignore.case = F) ~ "Influencer",
grepl("@kimberly.loaiza", account_name, ignore.case = F) ~ "Artista",
grepl("@kallmekris", account_name, ignore.case = F) ~ "Comédia",
grepl("@kylethomas", account_name, ignore.case = F) ~ "Influencer",
grepl("@kyliejenner", account_name, ignore.case = F) ~ "Influencer",
grepl("@juliamenugarcia", account_name, ignore.case = F) ~ "Influencer",
grepl("@daviddobrik", account_name, ignore.case = F) ~ "Influencer",
grepl("@adrilatinatv", account_name, ignore.case = F) ~ "Influencer",
grepl("@kikakiim", account_name, ignore.case = F) ~ "Influencer",
grepl("@kingbach", account_name, ignore.case = F) ~ "Comédia",
grepl("@karolg", account_name, ignore.case = F) ~ "Artista",
grepl("@era_ays", account_name, ignore.case = F) ~ "Influencer",
grepl("@spencerx", account_name, ignore.case = F) ~ "Artista",
grepl("@thekiryalife", account_name, ignore.case = F) ~ "Atleta",
grepl("@donaldducc", account_name, ignore.case = F) ~ "Comédia",
grepl("@mr_faisu_07", account_name, ignore.case = F) ~ "Comédia",
grepl("@benjikrol", account_name, ignore.case = F) ~ "Influencer",
grepl("@camilo", account_name, ignore.case = F) ~ "Artista",
grepl("@sameeksha_sud", account_name, ignore.case = F) ~ "Artista",
grepl("@topperguild", account_name, ignore.case = F) ~ "Comédia",
grepl("@psg", account_name, ignore.case = F) ~ "Institucional",
grepl("@xoteam", account_name, ignore.case = F) ~ "Influencer",
grepl("@scottsreality", account_name, ignore.case = F) ~ "Comida",
grepl("@nishaguragain", account_name, ignore.case = F) ~ "Influencer",
grepl("@awezdarbar", account_name, ignore.case = F) ~ "Artista",
grepl("@kodyantle", account_name, ignore.case = F) ~ "Animais",
grepl("@riyaz.14", account_name, ignore.case = F) ~ "Influencer",
grepl("@tirullipa", account_name, ignore.case = F) ~ "Comédia",
grepl("@wigofellas", account_name, ignore.case = F) ~ "Comédia",
grepl("@arianagrande", account_name, ignore.case = F) ~ "Artista"))
# Classifica nacionalidade das contas
tt_data <- tt_data %>%
mutate(nacionalidade = case_when(grepl("@happydogs2021", account_name, ignore.case = F) ~ "NA",
grepl("@avani", account_name, ignore.case = F) ~ "EUA",
grepl("@tiktok", account_name, ignore.case = F) ~ "China",
grepl("@anokhinalz", account_name, ignore.case = F) ~ "Rússia",
grepl("@babyariel", account_name, ignore.case = F) ~ "EUA",
grepl("@bellapoarch", account_name, ignore.case = F) ~ "Filipinas",
grepl("@brentrivera", account_name, ignore.case = F) ~ "EUA",
grepl("@charlidamelio", account_name, ignore.case = F) ~ "EUA",
grepl("@briandadeyanara", account_name, ignore.case = F) ~ "México",
grepl("@bts_official_bighit", account_name, ignore.case = F) ~ "Coreia do Sul",
grepl("@carlosferiag", account_name, ignore.case = F) ~ "Colômbia",
grepl("@cznburak", account_name, ignore.case = F) ~ "Turquia",
grepl("@darianrojasc", account_name, ignore.case = F) ~ "México",
grepl("@dixiedamelio", account_name, ignore.case = F) ~ "EUA",
grepl("@dobretwins", account_name, ignore.case = F) ~ "EUA",
grepl("@domelipa", account_name, ignore.case = F) ~ "México",
grepl("@elrodcontreras", account_name, ignore.case = F) ~ "México",
grepl("@gilmhercroes", account_name, ignore.case = F) ~ "EUA",
grepl("@homm9k", account_name, ignore.case = F) ~ "Cazaquistão",
grepl("@gordonramsayofficial", account_name, ignore.case = F) ~ "Escócia",
grepl("@itsjojosiwa", account_name, ignore.case = F) ~ "EUA",
grepl("@jamescharles", account_name, ignore.case = F) ~ "EUA",
grepl("@junya1gou", account_name, ignore.case = F) ~ "Japão",
grepl("@jasonderulo", account_name, ignore.case = F) ~ "EUA",
grepl("@joealbanese", account_name, ignore.case = F) ~ "EUA",
grepl("@khaby.lame", account_name, ignore.case = F) ~ "Senegal",
grepl("@addisonre", account_name, ignore.case = F) ~ "EUA",
grepl("@justmaiko", account_name, ignore.case = F) ~ "EUA",
grepl("@kimberly.loaiza", account_name, ignore.case = F) ~ "México",
grepl("@kallmekris", account_name, ignore.case = F) ~ "Canadá",
grepl("@kylethomas", account_name, ignore.case = F) ~ "Inglaterra",
grepl("@kyliejenner", account_name, ignore.case = F) ~ "EUA",
grepl("@juliamenugarcia", account_name, ignore.case = F) ~ "Espanha",
grepl("@daviddobrik", account_name, ignore.case = F) ~ "Eslováquia",
grepl("@adrilatinatv", account_name, ignore.case = F) ~ "Colômbia",
grepl("@kikakiim", account_name, ignore.case = F) ~ "Cazaquistão",
grepl("@kingbach", account_name, ignore.case = F) ~ "Canadá",
grepl("@karolg", account_name, ignore.case = F) ~ "Colômbia",
grepl("@era_ays", account_name, ignore.case = F) ~ "Rússia",
grepl("@spencerx", account_name, ignore.case = F) ~ "EUA",
grepl("@thekiryalife", account_name, ignore.case = F) ~ "Rússia",
grepl("@donaldducc", account_name, ignore.case = F) ~ "EUA",
grepl("@mr_faisu_07", account_name, ignore.case = F) ~ "Índia",
grepl("@benjikrol", account_name, ignore.case = F) ~ "Brasil",
grepl("@camilo", account_name, ignore.case = F) ~ "Colômbia",
grepl("@sameeksha_sud", account_name, ignore.case = F) ~ "Índia",
grepl("@topperguild", account_name, ignore.case = F) ~ "EUA",
grepl("@psg", account_name, ignore.case = F) ~ "França",
grepl("@xoteam", account_name, ignore.case = F) ~ "Rússia",
grepl("@scottsreality", account_name, ignore.case = F) ~ "Holanda",
grepl("@nishaguragain", account_name, ignore.case = F) ~ "Índia",
grepl("@awezdarbar", account_name, ignore.case = F) ~ "Índia",
grepl("@kodyantle", account_name, ignore.case = F) ~ "EUA",
grepl("@riyaz.14", account_name, ignore.case = F) ~ "Butão",
grepl("@tirullipa", account_name, ignore.case = F) ~ "Brasil",
grepl("@wigofellas", account_name, ignore.case = F) ~ "EUA",
grepl("@arianagrande", account_name, ignore.case = F) ~ "EUA"))
# PLOTANDO -------------------------------------------------------------------
# TIKTOKS - COMMENTS x LIKES
tt_data %>%
ggplot() +
aes(x=comment_count, y=digg_count, label=account_name, color=categoria, text=description) +
geom_point() -> p
ggplotly(p)
# CONTAS - COMMENTS x LIKES
tt_data %>%
group_by(account_name, categoria) %>%
summarise(likes=sum(digg_count), comments=sum(comment_count)) %>%
ggplot() +
aes(x=comments, y=likes, label=account_name, color=categoria) +
geom_text_repel(max.overlaps=15) +
geom_point() -> p
ggplotly(p)
# CATEGORIAS E CONTAS - LIKES
tt_data %>%
group_by(account_name,categoria) %>%
summarise(likes=sum(digg_count)) %>%
ggplot() +
aes(x=reorder(categoria,likes), y=likes, color=categoria, label=account_name) +
geom_point() +
geom_text_repel(max.overlaps=10) +
coord_flip() +
theme_minimal() -> p
ggplotly(p)
# CATEGORIAS - LIKES
tt_data %>%
group_by(account_name, categoria) %>%
summarise(likes=sum(digg_count)) %>%
ggplot() +
aes(x=reorder(categoria,likes, FUN=sum), y=likes, fill=reorder(account_name,likes,FUN=sum)) +
geom_col() +
coord_flip() +
theme_minimal()
# CATEGORIAS E NACIONALIDADES
tt_data %>%
group_by(account_name, categoria, nacionalidade) %>%
summarise(count = n()) %>%
group_by(categoria, nacionalidade) %>%
summarise(count = n()) %>%
ggplot() +
aes(x=reorder(nacionalidade, count, FUN=sum), y=count, fill=categoria) +
geom_col() +
coord_flip()
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment