install.packages("dplyr") install.packages("ggplot2") library(dplyr) library(ggplot2) # 변수 설정(조회수 분석 및 two samples test용) csvKR <- read.csv("KR_youtube_trending_data.csv",encoding='utf-8') csvUS <- read.csv("US_youtube_trending_data.csv",encoding='utf-8') # 변수 설정(이외 조회수 분석용) csvBR <- read.csv("BR_youtube_trending_data.csv",encoding='utf-8') csvCA <- read.csv("CA_youtube_trending_data.csv",encoding='utf-8') csvDE <- read.csv("DE_youtube_trending_data.csv",encoding='utf-8') csvFR <- read.csv("FR_youtube_trending_data.csv",encoding='utf-8') csvGB <- read.csv("GB_youtube_trending_data.csv",encoding='utf-8') csvIN <- read.csv("IN_youtube_trending_data.csv",encoding='utf-8') csvJP <- read.csv("JP_youtube_trending_data.csv",encoding='utf-8') csvMX <- read.csv("MX_youtube_trending_data.csv",encoding='utf-8') csvRU <- read.csv("RU_youtube_trending_data.csv",encoding='utf-8') csvUS <- read.csv("US_youtube_trending_data.csv",encoding='utf-8') # 영상별 최신 정보(조회수가 가장 많은 정보)만으로 추출 csvKR <- csvKR %>% arrange(trending_date) %>% group_by(video_id)%>% summarise(trending_date=last(trending_date), title=last(title), view_count=last(view_count), likes=last(likes)) %>% distinct() csvUS <- csvUS %>% arrange(trending_date) %>% group_by(video_id) %>% summarise(trending_date=last(trending_date), title=last(title), view_count=last(view_count), likes=last(likes)) %>% distinct() csvBR <- csvBR %>% arrange(trending_date) %>% group_by(video_id) %>% summarise(trending_date=last(trending_date), title=last(title), view_count=last(view_count), likes=last(likes)) %>% distinct() csvCA <- csvCA %>% arrange(trending_date) %>% group_by(video_id) %>% summarise(trending_date=last(trending_date), title=last(title), view_count=last(view_count), likes=last(likes)) %>% distinct() csvDE <- csvDE %>% arrange(trending_date) %>% group_by(video_id) %>% summarise(trending_date=last(trending_date), title=last(title), view_count=last(view_count), likes=last(likes)) %>% distinct() csvFR <- csvFR %>% arrange(trending_date) %>% group_by(video_id) %>% summarise(trending_date=last(trending_date), title=last(title), view_count=last(view_count), likes=last(likes)) %>% distinct() csvGB <- csvGB %>% arrange(trending_date) %>% group_by(video_id) %>% summarise(trending_date=last(trending_date), title=last(title), view_count=last(view_count), likes=last(likes)) %>% distinct() csvIN <- csvIN %>% arrange(trending_date) %>% group_by(video_id) %>% summarise(trending_date=last(trending_date), title=last(title), view_count=last(view_count), likes=last(likes)) %>% distinct() csvJP <- csvJP %>% arrange(trending_date) %>% group_by(video_id) %>% summarise(trending_date=last(trending_date), title=last(title), view_count=last(view_count), likes=last(likes)) %>% distinct() csvMX <- csvMX %>% arrange(trending_date) %>% group_by(video_id) %>% summarise(trending_date=last(trending_date), title=last(title), view_count=last(view_count), likes=last(likes)) %>% distinct() csvRU <- csvRU %>% arrange(trending_date) %>% group_by(video_id) %>% summarise(trending_date=last(trending_date), title=last(title), view_count=last(view_count), likes=last(likes)) %>% distinct() # 조회수가 없는 데이터 삭제 및 조회수 대비 좋아요 비율 컬럼 추가 csvKR <- csvKR %>% filter(view_count>0) %>% mutate(likes.per.views = likes/view_count*100) csvUS <- csvUS %>% filter(view_count>0) %>% mutate(likes.per.views = likes/view_count*100) csvBR <- csvBR %>% filter(view_count>0) %>% mutate(likes.per.views = likes/view_count*100) csvCA <- csvCA %>% filter(view_count>0) %>% mutate(likes.per.views = likes/view_count*100) csvDE <- csvDE %>% filter(view_count>0) %>% mutate(likes.per.views = likes/view_count*100) csvFR <- csvFR %>% filter(view_count>0) %>% mutate(likes.per.views = likes/view_count*100) csvGB <- csvGB %>% filter(view_count>0) %>% mutate(likes.per.views = likes/view_count*100) csvIN <- csvIN %>% filter(view_count>0) %>% mutate(likes.per.views = likes/view_count*100) csvJP <- csvJP %>% filter(view_count>0) %>% mutate(likes.per.views = likes/view_count*100) csvMX <- csvMX %>% filter(view_count>0) %>% mutate(likes.per.views = likes/view_count*100) csvRU <- csvRU %>% filter(view_count>0) %>% mutate(likes.per.views = likes/view_count*100) # 그래프에서의 색상별 분류를 위해 국가별 데이터베이스를 하나로 합치기 ## 각 국가별 데이터베이스에 국가 분류 추가 csvKR <- csvKR %>% mutate(trended_nation = "KR") csvUS <- csvUS %>% mutate(trended_nation = "US") csvBR <- csvBR %>% mutate(trended_nation = "BR") csvCA <- csvCA %>% mutate(trended_nation = "CA") csvDE <- csvDE %>% mutate(trended_nation = "DE") csvFR <- csvFR %>% mutate(trended_nation = "FR") csvGB <- csvGB %>% mutate(trended_nation = "GB") csvIN <- csvIN %>% mutate(trended_nation = "IN") csvJP <- csvJP %>% mutate(trended_nation = "JP") csvMX <- csvMX %>% mutate(trended_nation = "MX") csvRU <- csvRU %>% mutate(trended_nation = "RU") ## 하나로 합치기 infoKRUS <- rbind(csvKR, csvUS) info <- rbind(csvKR, csvUS) info <- rbind(info, csvBR) info <- rbind(info, csvCA) info <- rbind(info, csvDE) info <- rbind(info, csvFR) info <- rbind(info, csvGB) info <- rbind(info, csvIN) info <- rbind(info, csvJP) info <- rbind(info, csvMX) info <- rbind(info, csvRU) # 이상값 제거 전 한미 간 조회수 대비 좋아요 비율 비교 ## 그래프 프레임 제작 gg_krus_bo <- ggplot() + labs(title="\\"Likes/Views\\" Comparison in YouTube Trending Videos\\nbetween South Korea and the United States (Before Removing Outliers)", x="Likes/Views (%)",y="The Number of Videos", fill="Country", color="Country") + scale_x_continuous(breaks=seq(-0.25,35.25,1), labels = 0:35) + theme_bw() ## 데이터 추가 gg_krus_bo <- gg_krus_bo + geom_histogram(data=subset(infoKRUS, trended_nation == "KR"),binwidth = 0.5, aes(x=likes.per.views, fill=trended_nation, color=trended_nation), alpha=0.2) + geom_histogram(data=subset(infoKRUS, trended_nation == "US"),binwidth = 0.5, aes(x=likes.per.views, fill=trended_nation, color=trended_nation), alpha=0.2) ## 그래프 출력 gg_krus_bo # 국가별 데이터의 이상값 제거 후 하나로 합치기 ## 이상값 제거 csvKR2 <- csvKR %>% subset(likes.per.viewsboxplot(csvKR$likes.per.views)$stats[1,1]) %>% arrange(desc(likes.per.views)) csvUS2 <- csvUS %>% subset(likes.per.viewsboxplot(csvUS$likes.per.views)$stats[1,1]) %>% arrange(desc(likes.per.views)) csvBR2 <- csvBR %>% subset(likes.per.viewsboxplot(csvBR$likes.per.views)$stats[1,1]) %>% arrange(desc(likes.per.views)) csvCA2 <- csvCA %>% subset(likes.per.viewsboxplot(csvCA$likes.per.views)$stats[1,1]) %>% arrange(desc(likes.per.views)) csvDE2 <- csvDE %>% subset(likes.per.viewsboxplot(csvDE$likes.per.views)$stats[1,1]) %>% arrange(desc(likes.per.views)) csvFR2 <- csvFR %>% subset(likes.per.viewsboxplot(csvFR$likes.per.views)$stats[1,1]) %>% arrange(desc(likes.per.views)) csvGB2 <- csvGB %>% subset(likes.per.viewsboxplot(csvGB$likes.per.views)$stats[1,1]) %>% arrange(desc(likes.per.views)) csvIN2 <- csvIN %>% subset(likes.per.viewsboxplot(csvIN$likes.per.views)$stats[1,1]) %>% arrange(desc(likes.per.views)) csvJP2 <- csvJP %>% subset(likes.per.viewsboxplot(csvJP$likes.per.views)$stats[1,1]) %>% arrange(desc(likes.per.views)) csvMX2 <- csvMX %>% subset(likes.per.viewsboxplot(csvMX$likes.per.views)$stats[1,1]) %>% arrange(desc(likes.per.views)) csvRU2 <- csvRU %>% subset(likes.per.viewsboxplot(csvRU$likes.per.views)$stats[1,1]) %>% arrange(desc(likes.per.views)) ## 하나로 합치기 infoKRUS2 <- rbind(csvKR2, csvUS2) info2 <- rbind(csvKR2, csvUS2) info2 <- rbind(info2, csvBR2) info2 <- rbind(info2, csvCA2) info2 <- rbind(info2, csvDE2) info2 <- rbind(info2, csvFR2) info2 <- rbind(info2, csvGB2) info2 <- rbind(info2, csvIN2) info2 <- rbind(info2, csvJP2) info2 <- rbind(info2, csvMX2) info2 <- rbind(info2, csvRU2) # 이상값 제거 후 한미 간 조회수 대비 좋아요 비율 비교 ## 그래프 프레임 제작 gg_krus_ao <- ggplot() + labs(title="\\"Likes/Views\\" Comparison in YouTube Trending Videos\\nbetween South Korea and the United States (After Removing Outliers)", x="Likes/Views (%)",y="The Number of Videos", fill="Country", color="Country") + scale_x_continuous(breaks=seq(-0.25,35.25,1), labels = 0:35) + theme_bw() ## 데이터 추가 gg_krus_ao <- gg_krus_ao + geom_histogram(data=subset(infoKRUS2, trended_nation == "KR"),binwidth = 0.5, aes(x=likes.per.views, fill=trended_nation, color=trended_nation), alpha=0.2) + geom_histogram(data=subset(infoKRUS2, trended_nation == "US"),binwidth = 0.5, aes(x=likes.per.views, fill=trended_nation, color=trended_nation), alpha=0.2) ## 그래프 출력 gg_krus_ao # Test Statistics 실행 test <- t.test(csvKR2$likes.per.views, csvUS2$likes.per.views, alternative="less",conf.level = 0.95) test # critical value값 출력 qt(0.05,df=test$parameter) # YouTube Trending 운영 국가 country_list = c("BR", "CA", "DE", "FR", "GB", "IN", "JP", "KR", "MX", "RU", "US") # 이상값 제거 전 국가 간 조회수 대비 좋아요 비율 비교 ## 그래프 프레임 제작 gg_global_bo <- ggplot() + labs(title="\\"Likes/Views\\" Comparison in YouTube Trending Videos\\nbetween 11 countries (Before Removing Outliers)", x="Likes/Views (%)",y="The Number of Videos", fill="Country", color="Country") + scale_x_continuous(breaks=seq(0,90,10), labels = (0:9)*10) + theme_bw() + scale_fill_brewer(palette="Set3") + scale_color_brewer(palette="Set3") ## 국가별 데이터 추가 for (i in 1:length(country_list)){ gg_global_bo <- gg_global_bo + geom_density(data=subset(info, trended_nation == country_list[i]), size=1, aes(x=likes.per.views, fill=trended_nation, color=trended_nation), alpha=0)} ## 그래프 출력 gg_global_bo # 이상값 제거 후 국가 간 조회수 대비 좋아요 비율 비교 ## 그래프 프레임 제작 gg_global_ao <- ggplot() + labs(title="\\"Likes/Views\\" Comparison in YouTube Trending Videos\\nbetween 11 countries (After Removing Outliers)", x="Likes/Views (%)",y="The Number of Videos", fill="Country", color="Country") + scale_x_continuous(breaks=seq(0,27,1), labels = 0:27) + theme_bw() + scale_fill_brewer(palette="Set3") + scale_color_brewer(palette="Set3") + geom_line(size=1) ## 국가별 데이터 추가 for (i in 1:length(country_list)){ gg_global_ao <- gg_global_ao + geom_density(data=subset(info2, trended_nation == country_list[i]), size=1, aes(x=likes.per.views, fill=trended_nation, color=trended_nation), alpha=0)} ## 그래프 출력 gg_global_ao

[R] 우리나라 YouTube 시청자는 정말 '좋아요'를 잘 안 누르는가?