데이터 과학자와 함께 하는 제19대 대통령 선거

페이스북 페이지 활동성

민주당 문재인 후보와 국민의당 안철수 후보가 양강 체계를 구축하고 치열한 각축전을 벌이고 있다. 페이스북 페이지 활동성(CTR)을 비교한다.

페이스북 페이지 활동성 데이터

문재인, 안철수, 심상정 후보 페이스북 담벼락에 쓴 포스트(글, 동영상, 사진, 링크)를 긁어와서 활동성(CTR) 평가를 위한 데이터 형태로 준비한다.

# 0. 참고문헌 ---------------------------------------------------------------------
# library(Rfacebook)
# library(RCurl)
# library(tidyverse)
# library(lubridate)
# library(ggplot2)
# library(stringr)
# library(DT)
# library(gridExtra)
# library(ggthemes)
# library(extrafont)
# library(plotly)
# loadfonts()

fb_keys <- "EAACEdEose0cBANDQoParhyI0hNdNx9FBw5UPASmBVxb3UADLZA7aCDb4PZBOqZCJLkwXUyGVXv8oWyE0lF4AKjxqq0jo2Wr1HJn6CqXU6tQPFxozEbWsnLpnjZBoEzT4E1gamCMULLb42aTby5SlqdIvk0GIdIx3YajHYxtoZAaZC86cxgm44t"

# 1. 데이터 불러오기 ---------------------------------------------------------------------

## 1.1. 페북 데이터 데이터---------------------------------

start_date <- "2017/01/01"
end_date   <- "2017/05/02"

# 페북 포스트글 긁어오기: 일별 100개
get_fb_posts <- function(page_name, start_date, end_date, fb_keys) {
  # 긁어올 날짜 생성
  scraping_days <- seq(from=as.Date(start_date), to=as.Date(end_date), by='days')
  posts  <- data.frame()
  for(scrape_day in scraping_days){
    daily_post  <- data.frame()
    tryCatch({
      daily_post <- getPage(page=page_name, token=fb_keys,
                            since=as.Date(scrape_day, origin='1970-01-01'),
                            until=as.Date(scrape_day, origin='1970-01-01')+1)},
      error = function(e){})
    posts <- rbind(posts, daily_post)
  }
  return(posts)
}

# moon_df <- get_fb_posts("moonbyun1", start_date, end_date, fb_keys)
# ahn_df <- get_fb_posts("ahncs111", start_date, end_date, fb_keys)
# sim_df <- get_fb_posts("simsangjung", start_date, end_date, fb_keys)

moon_df <- read_csv("data/fb_moon_ctr.csv")
ahn_df  <- read_csv("data/fb_ahn_ctr.csv")
sim_df  <- read_csv("data/fb_sim_ctr.csv")

# 2. 데이터 변환 ---------------------------------------------------------------------
## 2.1. 안철수 -----------------------------------------------------------------------
ahn_time_df <- ahn_df %>% 
  dplyr::select(from_name, fb_date=created_time, type, likes_count, comments_count, shares_count) %>% 
  mutate(wday = week(fb_date))

ahn_ctr_df <- ahn_time_df %>% group_by(wday) %>% 
  summarise(posts = n(),
            likes_mean = mean(likes_count)) %>% 
  mutate(candid = "안철수")

ahn_ctr_tbl_df <- ahn_time_df %>% group_by(wday, type) %>% 
  summarise(posts = n(),
            likes_mean = mean(likes_count)) %>% 
  mutate(candid = "안철수")

## 2.2. 문재인 -----------------------------------------------------------------------
moon_time_df <- moon_df %>% 
  dplyr::select(from_name, fb_date=created_time, type, likes_count, comments_count, shares_count) %>% 
  mutate(wday = week(fb_date))

moon_ctr_df <- moon_time_df %>% group_by(wday) %>% 
  summarise(posts = n(),
            likes_mean = mean(likes_count)) %>% 
  mutate(candid = "문재인")

moon_ctr_tbl_df <- moon_time_df %>% group_by(wday, type) %>% 
  summarise(posts = n(),
            likes_mean = mean(likes_count)) %>% 
  mutate(candid = "문재인")

## 2.3. 심상정 -----------------------------------------------------------------------
sim_time_df <- sim_df %>% 
  dplyr::select(from_name, fb_date=created_time, type, likes_count, comments_count, shares_count) %>% 
  mutate(wday = week(fb_date))

sim_ctr_df <- sim_time_df %>% group_by(wday) %>% 
  summarise(posts = n(),
            likes_mean = mean(likes_count)) %>% 
  mutate(candid = "심상정")

sim_ctr_tbl_df <- sim_time_df %>% group_by(wday, type) %>% 
  summarise(posts = n(),
            likes_mean = mean(likes_count)) %>% 
  mutate(candid = "심상정")

## 2.3. 데이터 병합 -----------------------------------------------------------------------

fb_ctr_df <- bind_rows(ahn_ctr_df, moon_ctr_df) %>% 
  bind_rows(sim_ctr_df)

페이스북 페이지 활동성 시각화

17년 1월 첫주부터 최근까지 주별 활동성(좋아요 평균)을 각 후보별로 비교한다.

# 3. 시각화 -----------------------------------------------------------------------
## 3.1. 문재인, 안철수, 심상정 합쳐서 --------------------------------------------------
ctr_p <- ggplot(data=fb_ctr_df)  +
  aes(x=wday, y=likes_mean, color=candid) +
  geom_point(aes(colour=candid), alpha=0.9) +
  geom_line(aes(colour=candid), alpha=0.9) +
  scale_y_continuous(labels = scales::comma) +
  theme_tufte(base_family="NanumGothic") +
  theme(legend.position="none", 
        legend.title = element_blank(),
        plot.caption=element_text(hjust=0,size=8),
        plot.subtitle=element_text(face="bold"),
        axis.text=element_text(size=9.5))+
  labs(x="주간 통계: '17년 1월 1주차 ~",y="",title="",
       caption="\n source: https://developers.facebook.com/tools/explorer",
       subtitle="페이스북 주별 담벼락글 별 평균 좋아요") +
  scale_color_manual(values=c("#4286f4", "#1cd313", "#f2ee09"))

ggplotly(ctr_p)

## 3.2. 문재인, 안철수, 심상정 나눠서 --------------------------------------------------

ahn_p <- ggplot(data=fb_ctr_df %>% dplyr::filter(candid == "안철수"))  +
  aes(x=wday, y=likes_mean) +
  geom_point(aes(colour="green"), alpha=1.0, size=1.5) +
  geom_line(aes(colour="green"), alpha=0.7, size=0.7) +
  scale_y_continuous(labels = scales::comma) +
  theme_tufte(base_family="NanumGothic") +
  theme(legend.position="none", 
        legend.title = element_blank(),
        plot.caption=element_text(hjust=0,size=8),
        plot.subtitle=element_text(face="bold"),
        axis.text=element_text(size=9.5))+
  labs(x="주간 통계: '17년 1월 1주차 ~",y="",title="안철수 페북 페이지",
       caption="\n source: https://developers.facebook.com/tools/explorer",
       subtitle="페이스북 주별 담벼락글 별 평균 좋아요") +
  scale_color_manual(values=c("#1cd313"))

moon_p <- ggplot(data=fb_ctr_df %>% dplyr::filter(candid == "문재인"))  +
  aes(x=wday, y=likes_mean) +
  geom_point(aes(colour="blue"), alpha=1.0, size=1.5) +
  geom_line(aes(colour="blue"), alpha=0.7, size=0.7) +
  scale_y_continuous(labels = scales::comma) +
  theme_tufte(base_family="NanumGothic") +
  theme(legend.position="none", 
        legend.title = element_blank(),
        plot.caption=element_text(hjust=0,size=8),
        plot.subtitle=element_text(face="bold"),
        axis.text=element_text(size=9.5))+
  labs(x="주간 통계: '17년 1월 1주차 ~",y="",title="문재인 페북 페이지",
       caption="\n source: https://developers.facebook.com/tools/explorer",
       subtitle="페이스북 주별 담벼락글 별 평균 좋아요") +
  scale_color_manual(values=c("#4286f4"))

sim_p <- ggplot(data=fb_ctr_df %>% dplyr::filter(candid == "심상정"))  +
  aes(x=wday, y=likes_mean) +
  geom_point(aes(colour="yellow"), alpha=1.0, size=1.5) +
  geom_line(aes(colour="yellow"), alpha=0.7, size=0.7) +
  scale_y_continuous(labels = scales::comma) +
  theme_tufte(base_family="NanumGothic") +
  theme(legend.position="none", 
        legend.title = element_blank(),
        plot.caption=element_text(hjust=0,size=8),
        plot.subtitle=element_text(face="bold"),
        axis.text=element_text(size=9.5))+
  labs(x="주간 통계: '17년 1월 1주차 ~",y="",title="심상정 페북 페이지",
       caption="\n source: https://developers.facebook.com/tools/explorer",
       subtitle="페이스북 주별 담벼락글 별 평균 좋아요") +
  scale_color_manual(values=c("#f2ee09"))

grid.arrange(moon_p, sim_p, ahn_p, nrow=1)

페이스북 페이지 활동성 상세 집계표

안철수

DT::datatable(ahn_df %>% dplyr::select(from_name, id, type, likes=likes_count, comments=comments_count, 
                                       shares=shares_count, link))

DT::datatable(ahn_ctr_tbl_df %>% dplyr::select(candid, type, wday, posts, likes_mean)) %>% 
  DT::formatCurrency("likes_mean", "", interval=3, mark=',', digits=1)

문재인

DT::datatable(moon_df %>% dplyr::select(from_name, id, type, likes=likes_count, comments=comments_count, 
                                       shares=shares_count, link))

DT::datatable(moon_ctr_tbl_df %>% dplyr::select(candid, type, wday, posts, likes_mean)) %>% 
  DT::formatCurrency("likes_mean", "", interval=3, mark=',', digits=1)

심상정

DT::datatable(sim_df %>% dplyr::select(from_name, id, type, likes=likes_count, comments=comments_count, 
                                       shares=shares_count, link))

DT::datatable(sim_ctr_tbl_df %>% dplyr::select(candid, type, wday, posts, likes_mean)) %>% 
  DT::formatCurrency("likes_mean", "", interval=3, mark=',', digits=1)