Does International Collaboration Impact IMDB Ratings? (A Focus on Netflix)
Jeesung Ahn
03-03-2023
Executive Background & Summary
International collaborations in filmmaking offer the potential to make content more accessible to a global audience. These partnerships can introduce diverse perspectives to the production process and attract additional capital and investment. But do these advantages lead to higher viewer ratings? This study examines whether international collaboration on a single title is associated with higher IMDB ratings. Moreover, it explores the relationship between revenue and IMDB ratings, recognizing the influence of other factors, such as the COVID-19 pandemic, which increased demand for streaming services. I also focused specifically on Netflix Original movies, analyzing how their ratings have evolved as Netflix’s global subscriber base has expanded, and whether international collaboration correlates with higher ratings.
IMDB ratings typically average around 6.8, with a range from 1 to 10.
More International Collaboration:
Higher IMDB Ratings: Both in general and for Netflix Originals, international collaborations are associated with higher ratings.
More IMDB Votes: Internationally collaborative films receive more votes on IMDB, which can serve as a proxy for viewer counts, indicating broader global appeal.
Netflix Originals Are Receiving Lower Ratings: Although Netflix Originals are receiving lower ratings, this trend does not necessarily correlate with revenue or profit. Further research is needed to explore the underlying causes.
Recommendations for Further Research: More data, such as user interviews and reviews, is needed to pinpoint the factors driving these effects.
Dataset
- Netflix Data: Sourced from Kaggle
- IMDB Data: Sourced from IMDB Interfaces
- Netflix Original Data (as of 06.01.2021): Sourced from Kaggle
- Financial Statistics: Sourced from Business of Apps
Caveat: All data is based on titles available on Netflix as of mid-2021.
# Load data
netflix<-read.csv('data/netflix_movies_shows_titles.csv')
netflix_originals<-read.csv('data/NetflixOriginals.csv')
netflix_originals=merge(netflix_originals,netflix %>% select(Title,country),by=c('Title'))
imdb_titles<-read.csv('data/imdb.title.basics.csv')
imdb_ratings<-read.csv('data/imdb.title.ratings.csv')
# Clean data
imdb_titles['startYear'][imdb_titles['startYear']=='\\N']<-NA
imdb_titles['endYear'][imdb_titles['endYear']=='\\N']<-NA
imdb_titles['runtimeMinutes'][imdb_titles['runtimeMinutes']=='\\N']<-NA
imdb_titles=imdb_titles %>% mutate(startYear=as.numeric(startYear),endYear=as.numeric(endYear),runtimeMinutes=as.numeric(runtimeMinutes))
imdb_df<-merge(imdb_titles,imdb_ratings, by=c('tconst'))
yr=c()
for (i in 1:dim(netflix['date_added'])[1]){yr[i]=as.numeric(str_split_i(netflix['date_added'][i,],"-",1))}
netflix=netflix %>% mutate(year_added_to_netflix=yr) # add a 'year added to netflix' column
netflix['country'][netflix['country']=='']<-NA # blank to NA
yr=c()
for (i in 1:dim(netflix_originals['Premiere'])[1]){yr[i]=as.numeric(str_split_i(netflix_originals['Premiere'][i,],"-",1))}
netflix_originals=netflix_originals %>% mutate(premiere_year=yr) # add a 'year added to netflix' column
netflix_originals['country'][netflix_originals['country']=='']<-NA # blank to NA
# merge netflix titles with imdb ratings
netflix_ratings<-merge(netflix, imdb_df,by.x=c('Title','release_year'),by.y=c('primaryTitle','startYear'))
netflix_ratings=netflix_ratings %>% distinct(Title,director,release_year,.keep_all=TRUE) # remove duplicates
# sum(is.na(netflix_ratings$country)) # check the number of NAs (missing country data) N=118Netflix: Where Did the Movies/Shows Come From? (Ranked)
Movies and shows on Netflix were produced by 122 different countries.
A total of 8,807 movies/shows are included in this analysis.
Number of Contributing Countries per Title:
- 1 Country: 7,487 titles
- 2 Countries: 873 titles
- 3 Countries: 273 titles
- 4 Countries: 114 titles
- 5 Countries: 37 titles
- 6+ Countries: 23 titles (maximum number: 12)
Titles Produced by More Than 5 Countries:
- 6 Countries: 14 titles
- 7 Countries: 5 titles
- 8 Countries: 2 titles
- 9 Countries: 0 titles
- 10 Countries: 1 title
- 11 Countries: 0 titles
- 12 Countries: 1 title
# sum(is.na(netflix$country)) # check the number of NAs (missing country data) N=831
# Extract country data
temp=netflix %>% mutate(country1=str_split_i(country,",",1),
country2=str_split_i(country,",",2),
country3=str_split_i(country,",",3),
country4=str_split_i(country,",",4),
country5=str_split_i(country,",",5),
country6=str_split_i(country,",",6),
country7=str_split_i(country,",",7),
country8=str_split_i(country,",",8),
country9=str_split_i(country,",",9),
country10=str_split_i(country,",",10),
country11=str_split_i(country,",",11),
country12=str_split_i(country,",",12)) %>%
select(Title,country1,country2,country3,country4,country5,country6,country7,country8,country9,country10,country11, country12,year_added_to_netflix,Genre)
netflix_long_df<-melt(temp,id=c("Title","year_added_to_netflix","Genre"),na.rm=T) %>% arrange(Title)
netflix_long_df = netflix_long_df %>% mutate(country=str_trim(value,side=c("left")))
# Data engineering
netflix_long_df[459,]['country']<-"United States"
netflix_long_df[2639,]['country']<-"United Kingdom"
# which(netflix_long_df['country']=='')
netflix_long_df<-netflix_long_df[-c(1748,1990,2620,2934,3571,8661,9548),] # remove blanks
# Rank which country produced most shows/movies on Neflix
# unique(netflix_long_df$country) # movies/shows on Netflix were produced by 122 different countries
country_count=netflix_long_df %>% count(country) %>% arrange(desc(n)) %>%
mutate(perc=round(`n`/sum(`n`),3)) %>% mutate(percentage = scales::percent(perc))
# netflix_long_df %>% group_by(country) %>% count() %>%
# ungroup() %>%
# mutate(perc=round(`n`/sum(`n`),2)) %>% mutate(percentage = scales::percent(perc)) %>% arrange(desc(perc))
plot_df=temp%>%
mutate(number_of_countries=ifelse(is.na(country2),1,
ifelse(is.na(country3),2,
ifelse(is.na(country4),3,
ifelse(is.na(country5),4,
ifelse(is.na(country6),5,6
))))))plot_df=plot_df %>% count(number_of_countries)
ggplot(plot_df, aes(x=number_of_countries,y=n,fill=n))+ geom_col(fill = "black") +
geom_text(aes(label=n,fontface="bold"), vjust = -0.5, size =3,color="#E50914",family = 'AppleGothic') + theme_bw() +theme_classic()+
xlab(c("Number of Countries"))+ylab(c("Count"))+
scale_x_discrete(limits=c(seq(1,6,1)))+
ggtitle("Number of Countries that Collaborated on a Movie/Show on Netflix")+
theme(plot.title = element_text(hjust = 0.5,vjust=2.5,face="bold",family = 'AppleGothic'),
axis.text.x = element_text(size=10, family = 'mono'),
axis.text.y = element_text(size=10, family = 'mono'),
axis.title.x = element_text(size=10,face="bold", family = 'mono',hjust=1),
axis.title.y = element_text(size=10, face="bold",family = 'mono',vjust=1,hjust=1))
country_count=country_count %>% mutate("Production Country"=country, "Number of Titles"=n,"Percentage (%)"=perc*100,"Rank"=row_number())
# plot distribution of countries in table
formattable(country_count %>% select("Rank","Production Country", "Number of Titles", "Percentage (%)"), align=c("c","r","r","r"),
list(`Number of Titles`=color_bar("#E50914",fun=proportion),`Percentage (%)`=color_bar("gray")),
table.attr = 'style="font-size: 18px; font-family: AppleGothic";\"') | Rank | Production Country | Number of Titles | Percentage (%) |
|---|---|---|---|
| 1 | United States | 3687 | 36.8 |
| 2 | India | 1046 | 10.4 |
| 3 | United Kingdom | 806 | 8.1 |
| 4 | Canada | 445 | 4.4 |
| 5 | France | 393 | 3.9 |
| 6 | Japan | 318 | 3.2 |
| 7 | Spain | 232 | 2.3 |
| 8 | South Korea | 231 | 2.3 |
| 9 | Germany | 226 | 2.3 |
| 10 | Mexico | 169 | 1.7 |
| 11 | China | 161 | 1.6 |
| 12 | Australia | 160 | 1.6 |
| 13 | Egypt | 117 | 1.2 |
| 14 | Turkey | 112 | 1.1 |
| 15 | Hong Kong | 105 | 1.0 |
| 16 | Nigeria | 103 | 1.0 |
| 17 | Italy | 100 | 1.0 |
| 18 | Brazil | 97 | 1.0 |
| 19 | Argentina | 91 | 0.9 |
| 20 | Belgium | 90 | 0.9 |
| 21 | Indonesia | 90 | 0.9 |
| 22 | Taiwan | 89 | 0.9 |
| 23 | Philippines | 83 | 0.8 |
| 24 | Thailand | 70 | 0.7 |
| 25 | South Africa | 62 | 0.6 |
| 26 | Colombia | 51 | 0.5 |
| 27 | Netherlands | 50 | 0.5 |
| 28 | Denmark | 48 | 0.5 |
| 29 | Ireland | 46 | 0.5 |
| 30 | Sweden | 42 | 0.4 |
| 31 | Poland | 41 | 0.4 |
| 32 | Singapore | 41 | 0.4 |
| 33 | United Arab Emirates | 37 | 0.4 |
| 34 | New Zealand | 33 | 0.3 |
| 35 | Lebanon | 31 | 0.3 |
| 36 | Israel | 30 | 0.3 |
| 37 | Norway | 30 | 0.3 |
| 38 | Chile | 29 | 0.3 |
| 39 | Russia | 27 | 0.3 |
| 40 | Malaysia | 26 | 0.3 |
| 41 | Pakistan | 24 | 0.2 |
| 42 | Czech Republic | 22 | 0.2 |
| 43 | Switzerland | 19 | 0.2 |
| 44 | Romania | 14 | 0.1 |
| 45 | Uruguay | 14 | 0.1 |
| 46 | Saudi Arabia | 13 | 0.1 |
| 47 | Austria | 12 | 0.1 |
| 48 | Luxembourg | 12 | 0.1 |
| 49 | Finland | 11 | 0.1 |
| 50 | Greece | 11 | 0.1 |
| 51 | Hungary | 11 | 0.1 |
| 52 | Iceland | 11 | 0.1 |
| 53 | Bulgaria | 10 | 0.1 |
| 54 | Peru | 10 | 0.1 |
| 55 | Qatar | 10 | 0.1 |
| 56 | Jordan | 9 | 0.1 |
| 57 | Kuwait | 8 | 0.1 |
| 58 | 7 | 0.1 | |
| 59 | Serbia | 7 | 0.1 |
| 60 | Cambodia | 6 | 0.1 |
| 61 | Kenya | 6 | 0.1 |
| 62 | Morocco | 6 | 0.1 |
| 63 | Portugal | 6 | 0.1 |
| 64 | Vietnam | 6 | 0.1 |
| 65 | Ghana | 5 | 0.0 |
| 66 | West Germany | 5 | 0.0 |
| 67 | Bangladesh | 4 | 0.0 |
| 68 | Croatia | 4 | 0.0 |
| 69 | Iran | 4 | 0.0 |
| 70 | Venezuela | 4 | 0.0 |
| 71 | Algeria | 3 | 0.0 |
| 72 | Malta | 3 | 0.0 |
| 73 | Senegal | 3 | 0.0 |
| 74 | Slovenia | 3 | 0.0 |
| 75 | Soviet Union | 3 | 0.0 |
| 76 | Syria | 3 | 0.0 |
| 77 | Ukraine | 3 | 0.0 |
| 78 | Zimbabwe | 3 | 0.0 |
| 79 | Cayman Islands | 2 | 0.0 |
| 80 | Georgia | 2 | 0.0 |
| 81 | Guatemala | 2 | 0.0 |
| 82 | Iraq | 2 | 0.0 |
| 83 | Mauritius | 2 | 0.0 |
| 84 | Namibia | 2 | 0.0 |
| 85 | Nepal | 2 | 0.0 |
| 86 | Afghanistan | 1 | 0.0 |
| 87 | Albania | 1 | 0.0 |
| 88 | Angola | 1 | 0.0 |
| 89 | Armenia | 1 | 0.0 |
| 90 | Azerbaijan | 1 | 0.0 |
| 91 | Bahamas | 1 | 0.0 |
| 92 | Belarus | 1 | 0.0 |
| 93 | Bermuda | 1 | 0.0 |
| 94 | Botswana | 1 | 0.0 |
| 95 | Burkina Faso | 1 | 0.0 |
| 96 | Cameroon | 1 | 0.0 |
| 97 | Cuba | 1 | 0.0 |
| 98 | Cyprus | 1 | 0.0 |
| 99 | Dominican Republic | 1 | 0.0 |
| 100 | East Germany | 1 | 0.0 |
| 101 | Ecuador | 1 | 0.0 |
| 102 | Ethiopia | 1 | 0.0 |
| 103 | Jamaica | 1 | 0.0 |
| 104 | Kazakhstan | 1 | 0.0 |
| 105 | Latvia | 1 | 0.0 |
| 106 | Liechtenstein | 1 | 0.0 |
| 107 | Lithuania | 1 | 0.0 |
| 108 | Malawi | 1 | 0.0 |
| 109 | Mongolia | 1 | 0.0 |
| 110 | Montenegro | 1 | 0.0 |
| 111 | Mozambique | 1 | 0.0 |
| 112 | Nicaragua | 1 | 0.0 |
| 113 | Palestine | 1 | 0.0 |
| 114 | Panama | 1 | 0.0 |
| 115 | Paraguay | 1 | 0.0 |
| 116 | Puerto Rico | 1 | 0.0 |
| 117 | Samoa | 1 | 0.0 |
| 118 | Slovakia | 1 | 0.0 |
| 119 | Somalia | 1 | 0.0 |
| 120 | Sri Lanka | 1 | 0.0 |
| 121 | Sudan | 1 | 0.0 |
| 122 | Uganda | 1 | 0.0 |
| 123 | Vatican City | 1 | 0.0 |
Netflix: Is International Collaboration Associated with Higher IMDB Ratings?
- The analysis includes 931 shows and movies, with data on country of production, IMDb ratings, and availability on Netflix.
- Results indicate that movies and shows with contributions from multiple countries tend to receive higher ratings on IMDB.
# Data preparation
# extract country information
temp=netflix_ratings %>% mutate(country1=str_split_i(country,",",1),
country2=str_split_i(country,",",2),
country3=str_split_i(country,",",3),
country4=str_split_i(country,",",4),
country5=str_split_i(country,",",5),
country6=str_split_i(country,",",6),
country7=str_split_i(country,",",7),
country8=str_split_i(country,",",8),country9=str_split_i(country,",",9))
# sum(is.na(temp$country)) # 118 titles have no country information
# sum(is.na(temp$country1)==FALSE) # 1799 movies/shows have 1 country involved
# sum(is.na(temp$country2)==FALSE) # 318 movies/shows have 2 countries involved
# sum(is.na(temp$country3)==FALSE) # 115 movies/shows have 3 countries involved
# sum(is.na(temp$country4)==FALSE) # 49 movies/shows have 4 countries involved
# sum(is.na(temp$country5)==FALSE) # 13 movies/shows have 5 countries involved
# sum(is.na(temp$country6)==FALSE) # 3 movies/shows have 6 countries involved
# sum(is.na(temp$country7)==FALSE) # 4 movies/shows have 7 countries involved
# sum(is.na(temp$country8)==FALSE) # 1 movies/shows have 8 countries involved
# sum(is.na(temp$country9)==FALSE) # 1 movies/shows have 12 countries involved (max number)
temp=temp %>%
select(Title,country,country1,country2,country3,country4,country5,country6,country7,country8,country9,
year_added_to_netflix,Genre,averageRating,numVotes)
nf_df_cleaned=temp%>%
mutate(number_of_countries=ifelse(is.na(country2),1,
ifelse(is.na(country3),2,
ifelse(is.na(country4),3,
ifelse(is.na(country5),4,
ifelse(is.na(country6),5,6)))))) # 6 means 6 or more countries
# U.S. vs. other countries (binary coding: other 0 vs. U.S. 1) (collab 1 vs. single country 0)
nf_df_cleaned=nf_df_cleaned %>% mutate(collab_or_not = ifelse(number_of_countries>=2,1,0),
us_or_not=ifelse(grepl('United States',country),1,0))Average Ratings
ggplot(data=nf_df_cleaned,aes(x=number_of_countries,y=averageRating))+
geom_smooth(method=lm,colour="#E50914",size=2,se=FALSE)+
scale_x_discrete(limits=c(seq(1,6,1))) +
theme_bw()+theme_classic()+
labs(x="Number of Countries", y="Average IMDb Ratings", title="International Collaborations Predict Higher Ratings on IMDb")+
theme(plot.title = element_text(hjust = 0.5,vjust=2.5,size=15,face="bold",family = 'AppleGothic'),
axis.text.x = element_text(size=10, family = 'mono'),
axis.text.y = element_text(size=10, family = 'mono',angle=45),
axis.title.x = element_text(size=10,face="bold", family = 'mono',hjust=1),
axis.title.y = element_text(size=10, face="bold",family = 'mono',vjust=1,hjust=1))
# summary(lm(averageRating~number_of_countries,data=nf_df_cleaned))
# qqnorm(x=nf_df_cleaned$number_of_countries,y=nf_df_cleaned$averageRating) normality
# cor.test(nf_df_cleaned$numVotes,nf_df_cleaned$averageRating) # titles that have more votes tend to score higher on ratingsNumber of Votes
ggplot(data=nf_df_cleaned,aes(x=number_of_countries,y=numVotes))+
geom_smooth(method=lm,colour="#E50914",size=2,se=FALSE)+
scale_x_discrete(limits=c(seq(1,6,1)))+ theme_bw()+theme_classic()+
labs(x="Number of Countries", y="Number of Votes", title="International Collaborations Predict More Votes on IMDb")+
theme(plot.title = element_text(hjust = 0.5,vjust=2.5,size=15,face="bold",family = 'AppleGothic'),
axis.text.x = element_text(size=10, family = 'mono'),
axis.text.y = element_text(size=10, family = 'mono',angle=45),
axis.title.x = element_text(size=10,face="bold", family = 'mono',hjust=1),
axis.title.y = element_text(size=10, face="bold",family = 'mono',vjust=1,hjust=1))
# summary(lm(log(numVotes)~log(number_of_countries),data=nf_df_cleaned)) # proxy for the number of viewers # log transformed for normality
# cor.test(nf_df_cleaned$numVotes,nf_df_cleaned$averageRating) # titles that have more votes tNetflix Originals: Where Did the Movies Come From?
How about Netflix Originals, specifically?
- Out of 504 movies analyzed:
- Netflix Original movies were produced by 43 different countries.
- 430 movies/shows involved a single country in their production.
- 59 movies/shows involved 2 countries.
- 14 movies/shows involved 3 countries.
- 1 movie/show involved 4 countries.
- The United States continues to dominate in the production of Netflix Originals.
# Data Preparation
# Extract country data
temp=netflix_originals %>% mutate(country1=str_split_i(country,",",1),
country2=str_split_i(country,",",2),
country3=str_split_i(country,",",3),
country4=str_split_i(country,",",4))
temp=temp %>% distinct(Title,Genre,Language,.keep_all=TRUE) # remove duplicates
temp1=temp %>%
select(Title,country1,country2,country3,country4,premiere_year,IMDB.Score,Genre)
netflix_originals_long_df<-melt(temp1,id=c("Title","premiere_year","IMDB.Score","Genre"),na.rm=T) %>% arrange(Title)
netflix_originals_long_df = netflix_originals_long_df %>% mutate(country=str_trim(value,side=c("left")))
# Rank which country produced most Netflix Original movies
# unique(netflix_originals_long_df$country) # Netflix original movies were produced by 43 different countries
country_count=netflix_originals_long_df %>% count(country) %>% arrange(desc(n)) %>%
mutate(perc=round(`n`/sum(`n`),3)) %>% mutate(percentage = scales::percent(perc))
# plot table
formattable(country_count %>% select(country,n,percentage), list(`n`=color_bar(color="lightblue")))| country | n | percentage |
|---|---|---|
| United States | 327 | 56.4% |
| India | 38 | 6.6% |
| United Kingdom | 38 | 6.6% |
| Canada | 17 | 2.9% |
| France | 16 | 2.8% |
| Italy | 16 | 2.8% |
| Spain | 14 | 2.4% |
| Mexico | 11 | 1.9% |
| Indonesia | 9 | 1.6% |
| Brazil | 8 | 1.4% |
| Germany | 8 | 1.4% |
| South Korea | 8 | 1.4% |
| Japan | 7 | 1.2% |
| Argentina | 6 | 1.0% |
| Hungary | 4 | 0.7% |
| South Africa | 4 | 0.7% |
| Turkey | 4 | 0.7% |
| Australia | 3 | 0.5% |
| Belgium | 3 | 0.5% |
| Netherlands | 3 | 0.5% |
| Philippines | 3 | 0.5% |
| Poland | 3 | 0.5% |
| Thailand | 3 | 0.5% |
| Austria | 2 | 0.3% |
| China | 2 | 0.3% |
| Denmark | 2 | 0.3% |
| Ireland | 2 | 0.3% |
| Nigeria | 2 | 0.3% |
| Norway | 2 | 0.3% |
| Sweden | 2 | 0.3% |
| 1 | 0.2% | |
| Albania | 1 | 0.2% |
| Cambodia | 1 | 0.2% |
| Chile | 1 | 0.2% |
| Greece | 1 | 0.2% |
| Iceland | 1 | 0.2% |
| Iran | 1 | 0.2% |
| Israel | 1 | 0.2% |
| Malaysia | 1 | 0.2% |
| Pakistan | 1 | 0.2% |
| Switzerland | 1 | 0.2% |
| Ukraine | 1 | 0.2% |
| United Arab Emirates | 1 | 0.2% |
nf_originals_df_cleaned=temp%>%
mutate(number_of_countries=ifelse(is.na(country2),1,
ifelse(is.na(country3),2,
ifelse(is.na(country4),3,4)))) # 6 means 6 or more countries
# U.S. vs. other countries (binary coding: other 0 vs. U.S. 1) (collab 1 vs. single country 0)
nf_originals_df_cleaned=nf_originals_df_cleaned %>% mutate(collab_or_not = ifelse(number_of_countries>=2,1,0),
us_or_not=ifelse(grepl('United States',country),1,0))
originals_country_count=netflix_originals_long_df %>% count(country) %>%
arrange(desc(n)) %>%mutate(perc=round(`n`/sum(`n`),3)) %>% mutate(percentage = scales::percent(perc))
originals_country_count=originals_country_count %>% mutate("Production Country"=country,
"Number of Movies/Shows"=n,"Percentage (%)"=perc*100 )plot_originals_df<-nf_originals_df_cleaned %>% count(number_of_countries)
ggplot(plot_originals_df, aes(x=number_of_countries,y=n,fill=n))+ geom_col(fill = "black") +
geom_text(aes(label=n,fontface="bold"), vjust = -0.5, size =3,color="#E50914",family = 'AppleGothic') + theme_bw() +theme_classic()+
xlab(c("Number of Countries"))+ylab(c("Count"))+
scale_x_discrete(limits=c(seq(1,4,1)))+
ggtitle("Number of Countries that Collaborated on a Netflix Original movie")+
theme(plot.title = element_text(hjust = 0.5,vjust=2.5,face="bold",family = 'AppleGothic'),
axis.text.x = element_text(size=10, family = 'mono'),
axis.text.y = element_text(size=10, family = 'mono'),
axis.title.x = element_text(size=10,face="bold", family = 'mono',hjust=1),
axis.title.y = element_text(size=10, face="bold",family = 'mono',vjust=3,hjust=1))
# plot distribution of countries in table
formattable(originals_country_count %>% select("Production Country", "Number of Movies/Shows", "Percentage (%)"), align=c("r","r","r"),
list(`Number of Movies/Shows`=color_bar("#E50914"),`Percentage (%)`=color_bar("gray"))) | Production Country | Number of Movies/Shows | Percentage (%) |
|---|---|---|
| United States | 327 | 56.4 |
| India | 38 | 6.6 |
| United Kingdom | 38 | 6.6 |
| Canada | 17 | 2.9 |
| France | 16 | 2.8 |
| Italy | 16 | 2.8 |
| Spain | 14 | 2.4 |
| Mexico | 11 | 1.9 |
| Indonesia | 9 | 1.6 |
| Brazil | 8 | 1.4 |
| Germany | 8 | 1.4 |
| South Korea | 8 | 1.4 |
| Japan | 7 | 1.2 |
| Argentina | 6 | 1.0 |
| Hungary | 4 | 0.7 |
| South Africa | 4 | 0.7 |
| Turkey | 4 | 0.7 |
| Australia | 3 | 0.5 |
| Belgium | 3 | 0.5 |
| Netherlands | 3 | 0.5 |
| Philippines | 3 | 0.5 |
| Poland | 3 | 0.5 |
| Thailand | 3 | 0.5 |
| Austria | 2 | 0.3 |
| China | 2 | 0.3 |
| Denmark | 2 | 0.3 |
| Ireland | 2 | 0.3 |
| Nigeria | 2 | 0.3 |
| Norway | 2 | 0.3 |
| Sweden | 2 | 0.3 |
| 1 | 0.2 | |
| Albania | 1 | 0.2 |
| Cambodia | 1 | 0.2 |
| Chile | 1 | 0.2 |
| Greece | 1 | 0.2 |
| Iceland | 1 | 0.2 |
| Iran | 1 | 0.2 |
| Israel | 1 | 0.2 |
| Malaysia | 1 | 0.2 |
| Pakistan | 1 | 0.2 |
| Switzerland | 1 | 0.2 |
| Ukraine | 1 | 0.2 |
| United Arab Emirates | 1 | 0.2 |
Netflix Originals: Is international collaboration associated with higher IMDB ratings?
- Movies and shows produced by multiple countries tend to receive significantly higher IMDB ratings.
Average Ratings
# summary(lm(IMDB.Score~number_of_countries,data=nf_originals_df_cleaned))
ggplot(data=nf_originals_df_cleaned,aes(x=number_of_countries,y=IMDB.Score))+
geom_smooth(method=lm,colour="#E50914",size=2,se=FALSE)+
scale_x_discrete(limits=c(seq(1,6,1))) + coord_cartesian(ylim=c(6,7))+
theme_bw()+theme_classic()+
labs(x="Number of Countries", y="Average IMDb Ratings", title="International Collaborations Predict Higher Ratings")+
theme(plot.title = element_text(hjust = 0.5,vjust=2.5,size=15,face="bold",family = 'AppleGothic'),
axis.text.x = element_text(size=10, family = 'mono'),
axis.text.y = element_text(size=10, family = 'mono',angle=45),
axis.title.x = element_text(size=10,face="bold", family = 'mono',hjust=1),
axis.title.y = element_text(size=10, face="bold",family = 'mono',vjust=1,hjust=1))
# nf_originals_df_cleaned %>% group_by(premiere_year) %>% summarise(average_n_of_countries=mean(number_of_countries))Overall Content on Netflix vs. Netflix Originals (IMDb Ratings)
# nf_originals_df_cleaned[which(nf_originals_df_cleaned$IMDB.Score==max(nf_originals_df_cleaned$IMDB.Score))[1],] # highest rating
# nf_originals_df_cleaned[which(nf_originals_df_cleaned$IMDB.Score==min(nf_originals_df_cleaned$IMDB.Score))[1],] # lowest rating
# t.test(nf_originals_df_cleaned$IMDB.Score,nf_df_cleaned$averageRating) # netflix originals have significantly lower ratings, compared to all contents available on Netflix combined
bar_df<-data.frame(group=c('Netflix Overall','Netflix Originals'),Average_Ratings=c(round(mean(nf_df_cleaned$averageRating),2),round(mean(nf_originals_df_cleaned$IMDB.Score),2)))
ggplot(bar_df, aes(x=group,y=Average_Ratings,fill=Average_Ratings))+
geom_col(fill = "black",width=0.5) +
coord_cartesian(ylim=c(5,7))+
geom_text(aes(label=Average_Ratings,fontface="bold"), vjust = -0.5, size =3,color="#E50914",family = 'AppleGothic') +
theme_bw() +theme_classic()+ylab(c("Average IMDb Ratings"))+
ggtitle("IMDb Ratings: Overall Content on Netflix vs. Netflix Originals")+
theme(plot.title = element_text(hjust = 0.5,vjust=2.5,face="bold",family = 'AppleGothic'),
axis.text.x = element_text(size=10, family = 'mono'),
axis.text.y = element_text(size=10, family = 'mono'),
axis.title.x = element_blank(),
axis.title.y = element_text(size=10, face="bold",family = 'mono',vjust=3,hjust=1))
Appendix: Are IMDb Ratings Correlated with Financial Statistics?
Despite growing revenue, increased spending on content, and a rising number of subscribers, IMDb ratings for Netflix Originals have been declining. Interestingly, while there is a statistically significant relationship observed, average IMDb ratings do not correlate directly with revenue. This may be due to a diverse audience base or other confounding factors, but the causality remains unclear.
- Next Steps
- Alternative Measures: Consider other metrics beyond IMDb ratings to evaluate user experience.
- Text Mining Reviews: Conduct text mining on user reviews to identify factors contributing to the decline in average ratings.
# nf_originals_df_cleaned %>% group_by(premiere_year) %>% summarise(Mean_IMDb_Ratings=mean(IMDB.Score)) # netflix originals
# nf_df_cleaned %>% select(year_added_to_netflix,averageRating,numVotes,number_of_countries) %>% group_by(year_added_to_netflix) %>% summarise(Average_IMDb_Ratings=mean(averageRating)) # all content
usage_stats<-data.frame(year=c(2011,2012,2013,2014,2015,2016,2017,2018,2019,2020,2021,2022),
revenue_bn=c(3.1,3.5,4.3,5.4,6.7,8.8,11.6,15.7,20.1,24.9,29.6,31.6),
net_income=c(225,17,112,266,122,188,560,1211,1867,2761,5116,4491),
content_spend_bn=c(NA,NA,NA,NA,NA,6.88,8.91,12,13.9,11.8,17.7,16.8),
subscribers_mm=c(21.5,25.71,35.63,47.99,62.71,79.9,99.04,124.35,151.56,192.95,209,220.6),
netflix_originals_imdb=c(NA,NA,NA,6.4,6.88,6.34,6.29,6.25,6.22,6.10,6.04,NA),
overall_imdb=c(c(6.5,NA, 8.1,7.56,7.02, 6.75,6.72,6.8,6.72,6.70, 6.75,NA)))
imdb_average_ratings<-data.frame(year=c(2014,2015,2016,2017,2018,2019,2020,2021),
group=c(rep("Netflix Originals",8),rep("Overall on Netflix",8)),
ratings=c(c(6.4,6.88,6.34,6.29,6.25,6.22,6.10,6.04),c(7.56,7.02, 6.75,6.72,6.8,6.72,6.70,6.75)))
ggplot(data=imdb_average_ratings, aes(x=year,y=ratings,fill=group))+
geom_line(aes(color=group),size=2,arrow=arrow())+scale_x_discrete(limits=c(seq(2014,2021,1)))+ scale_color_manual(values=c("#E50914","black"))+
theme_bw() +
labs(x="Year", y="Average IMDb Rating", title="Yearly Trends in IMDb Ratings",color=NULL)+
theme(plot.title = element_text(hjust = 0.5,vjust=2.5,size=15,face="bold",family = 'AppleGothic'),
axis.text.x = element_text(size=10, family = 'mono'),
axis.text.y = element_text(size=10, family = 'mono'),
axis.title.x = element_text(size=10,face="bold", family = 'mono',hjust=1),
axis.title.y = element_text(size=10, face="bold",family = 'mono',vjust=1,hjust=1)) 
# nf_originals_df_cleaned %>% group_by(premiere_year) %>% summarise(Mean_IMDb_Ratings=mean(IMDB.Score)) # netflix originals
# nf_df_cleaned %>% select(year_added_to_netflix,averageRating,numVotes,number_of_countries) %>% group_by(year_added_to_netflix) %>% summarise(Average_IMDb_Ratings=mean(averageRating)) # all content
usage_stats<-data.frame(year=c(2011,2012,2013,2014,2015,2016,2017,2018,2019,2020,2021,2022),
revenue_bn=c(3.1,3.5,4.3,5.4,6.7,8.8,11.6,15.7,20.1,24.9,29.6,31.6),
net_income_bn=c(0.225,0.017,0.112,0.266,0.122,0.188,0.560,1.211,1.867,2.761,5.116,4.491),
content_spend_bn=c(NA,NA,NA,NA,NA,6.88,8.91,12,13.9,11.8,17.7,16.8),
subscribers_mm=c(21.5,25.71,35.63,47.99,62.71,79.9,99.04,124.35,151.56,192.95,209,220.6),
netflix_originals_imdb=c(NA,NA,NA,6.4,6.88,6.34,6.29,6.25,6.22,6.10,6.04,NA),
overall_imdb=c(c(6.5,NA, 8.1,7.56,7.02, 6.75,6.72,6.8,6.72,6.70, 6.75,NA)))
finance<-data.frame(year=c(2011,2012,2013,2014,2015,2016,2017,2018,2019,2020,2021,2022),
group=c(rep("Revenue ($B)",12),rep("Net Income ($B)",12),rep("Content Spend ($B)",12)),
money=c(c(3.1,3.5,4.3,5.4,6.7,8.8,11.6,15.7,20.1,24.9,29.6,31.6),c(0.225,0.017,0.112,0.266,0.122,0.188,0.560,1.211,1.867,2.761,5.116,4.491),c(NA,NA,NA,NA,NA,6.88,8.91,12,13.9,11.8,17.7,16.8)))
ggplot(data=finance, aes(x=year,y=money,fill=group))+
geom_line(aes(color=group),size=2,arrow=arrow())+scale_x_discrete(limits=c(seq(2011,2022,1)))+
theme_bw() +
labs(x="Year", y="Billions (USD)", title="Netflix financial Statistics",color=NULL)+
theme(plot.title = element_text(hjust = 0.5,vjust=2.5,size=15,face="bold",family = 'AppleGothic'),
axis.text.x = element_text(size=10, family = 'mono'),
axis.text.y = element_text(size=10, family = 'mono'),
axis.title.x = element_text(size=10,face="bold", family = 'mono',hjust=1),
axis.title.y = element_text(size=10, face="bold",family = 'mono',vjust=1,hjust=1)) 
# cor.test(usage_stats$revenue_bn,usage_stats$netflix_originals_imdb)
# cor.test(usage_stats$content_spend_bn,usage_stats$netflix_originals_imdb)
# cor.test(usage_stats$subscribers_mm,usage_stats$netflix_originals_imdb)
# cor.test(usage_stats$revenue_bn,usage_stats$overall_imdb)
# cor.test(usage_stats$content_spend_bn,usage_stats$overall_imdb)
# cor.test(usage_stats$subscribers_mm,usage_stats$overall_imdb)