Video Game Sales Worldwide
Leo Nakaya
2023-03-05
Video_Games <- read.csv("/Users/apple/Desktop/R/Video_Games.csv")
data(Video_Games)## Warning in data(Video_Games): data set 'Video_Games' not found#Introduction_and_Preparation
summary(Video_Games)## Name Platform Year_of_Release Genre
## Length:16719 Length:16719 Length:16719 Length:16719
## Class :character Class :character Class :character Class :character
## Mode :character Mode :character Mode :character Mode :character
##
##
##
##
## Publisher NA_Sales EU_Sales JP_Sales
## Length:16719 Min. : 0.0000 Min. : 0.000 Min. : 0.0000
## Class :character 1st Qu.: 0.0000 1st Qu.: 0.000 1st Qu.: 0.0000
## Mode :character Median : 0.0800 Median : 0.020 Median : 0.0000
## Mean : 0.2633 Mean : 0.145 Mean : 0.0776
## 3rd Qu.: 0.2400 3rd Qu.: 0.110 3rd Qu.: 0.0400
## Max. :41.3600 Max. :28.960 Max. :10.2200
##
## Other_Sales Global_Sales Critic_Score Critic_Count
## Min. : 0.00000 Min. : 0.0100 Min. :13.00 Min. : 3.00
## 1st Qu.: 0.00000 1st Qu.: 0.0600 1st Qu.:60.00 1st Qu.: 12.00
## Median : 0.01000 Median : 0.1700 Median :71.00 Median : 21.00
## Mean : 0.04733 Mean : 0.5335 Mean :68.97 Mean : 26.36
## 3rd Qu.: 0.03000 3rd Qu.: 0.4700 3rd Qu.:79.00 3rd Qu.: 36.00
## Max. :10.57000 Max. :82.5300 Max. :98.00 Max. :113.00
## NA's :8582 NA's :8582
## User_Score User_Count Developer Rating
## Length:16719 Min. : 4.0 Length:16719 Length:16719
## Class :character 1st Qu.: 10.0 Class :character Class :character
## Mode :character Median : 24.0 Mode :character Mode :character
## Mean : 162.2
## 3rd Qu.: 81.0
## Max. :10665.0
## NA's :9129library(ggplot2)
library(dplyr)##
## Attaching package: 'dplyr'## The following objects are masked from 'package:stats':
##
## filter, lag## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, unionVideo_Games$Year_of_Release = as.numeric(as.character(Video_Games$Year_of_Release)) ## Warning: NAs introduced by coercionVideo_Games$User_Score = as.numeric(as.character(Video_Games$User_Score))## Warning: NAs introduced by coercionsummary(Video_Games)## Name Platform Year_of_Release Genre
## Length:16719 Length:16719 Min. :1980 Length:16719
## Class :character Class :character 1st Qu.:2003 Class :character
## Mode :character Mode :character Median :2007 Mode :character
## Mean :2006
## 3rd Qu.:2010
## Max. :2020
## NA's :269
## Publisher NA_Sales EU_Sales JP_Sales
## Length:16719 Min. : 0.0000 Min. : 0.000 Min. : 0.0000
## Class :character 1st Qu.: 0.0000 1st Qu.: 0.000 1st Qu.: 0.0000
## Mode :character Median : 0.0800 Median : 0.020 Median : 0.0000
## Mean : 0.2633 Mean : 0.145 Mean : 0.0776
## 3rd Qu.: 0.2400 3rd Qu.: 0.110 3rd Qu.: 0.0400
## Max. :41.3600 Max. :28.960 Max. :10.2200
##
## Other_Sales Global_Sales Critic_Score Critic_Count
## Min. : 0.00000 Min. : 0.0100 Min. :13.00 Min. : 3.00
## 1st Qu.: 0.00000 1st Qu.: 0.0600 1st Qu.:60.00 1st Qu.: 12.00
## Median : 0.01000 Median : 0.1700 Median :71.00 Median : 21.00
## Mean : 0.04733 Mean : 0.5335 Mean :68.97 Mean : 26.36
## 3rd Qu.: 0.03000 3rd Qu.: 0.4700 3rd Qu.:79.00 3rd Qu.: 36.00
## Max. :10.57000 Max. :82.5300 Max. :98.00 Max. :113.00
## NA's :8582 NA's :8582
## User_Score User_Count Developer Rating
## Min. :0.000 Min. : 4.0 Length:16719 Length:16719
## 1st Qu.:6.400 1st Qu.: 10.0 Class :character Class :character
## Median :7.500 Median : 24.0 Mode :character Mode :character
## Mean :7.125 Mean : 162.2
## 3rd Qu.:8.200 3rd Qu.: 81.0
## Max. :9.700 Max. :10665.0
## NA's :9129 NA's :9129In the first stage, we imported this dataset named “Video Games Sales 2022” that was downloaded from kaggle. At first glance, in summary we noticed that the column 3 “Year_of_Release” and 13 “User_Score” are character type so we turned into numeric type for analysis purposes. Also, we installed needed libraries for this project. Hence, our data is almost ready to start to make some graphs.
#Adding_new_column
Video_Games <- Video_Games %>%
mutate(Year = case_when(Year_of_Release >= 2010 ~ "10s"
,Year_of_Release >= 2000 ~ "00s"
,Year_of_Release >= 1990 ~ "90s"
,Year_of_Release >= 1980 ~ "80s"))
Video_Games$Year <- factor(Video_Games$Year,
levels = c("80s","90s","00s","10s"))We proceeded adding a new category “Year”, classifying by the period that those games were released “80s”,“90s”,“00s”(noughties) and “10s”(The Tens). Also, we created a list in the order we wanted to avoid problems when sorting by year.
#Global_Sales_Histogram
ggplot()+geom_histogram(data=Video_Games,aes(x=Year_of_Release),
fill="gray",color="black",binwidth=1, na.rm = TRUE)+
labs(x="Year of Release",y="Total Amount of Copies Sold",
title="Global Sales by Year")+
theme(plot.title = element_text(hjust = 0.5))
We created a Histogram with our data to get a general perspective, realizing that from 1990 to 2010, overall sales increased exponentially.
#Top_10
Top_10 <- Video_Games %>%
filter(Global_Sales > 28.30)
dat <- data.frame(x=c(Top_10$Name),
y=c(Top_10$Global_Sales))
barplot(dat$y, names.arg=dat$x, cex.names=.5, las = 2,
main="The Top 10 Best-Selling Video Games of All Time")
Later, we wanted to know top best-selling video games, so we created a “Top 10” and found out many famous video games such as: Wii Sports with over 82 Million copies sold or Super Mario Bros. with 40.
#Top_20
Top_20 <- Video_Games %>%
filter(Global_Sales > 20.14)
dat <- data.frame(x=c(Top_20$Name),
y=c(Top_20$Global_Sales))
barplot(dat$y, names.arg=dat$x, cex.names=.5, las = 2,
main="The Top 20 Best-Selling Video Games of All Time")
After, we made a “Top 20” to find other well-known video games worldwide, in this case we recognized Grand Theft Auto and Pokemon series.
#Top_100
Top_100 <- Video_Games %>%
filter(Global_Sales > 7.38)
ggplot(Top_100, aes(x=Platform, y=Global_Sales,
fill=Platform))+
geom_boxplot(alpha=0.6)+
labs(x="Platform", y="Total Amount of Copies Sold",
title= "The Top 100 Best-Selling Video Games by Platform")+
theme(legend.position = "none")+
theme(panel.background = element_blank(),
panel.grid.major = element_blank(),
panel.grid.minor = element_blank())+
theme(plot.title = element_text(hjust = 0.5, face="bold"))
Since the “Top 100” represents over 90% of total sales, we made a boxplot with this top and researched which platforms got a huge part of those sales.
#Excluding_number1
Excluding_number1 <- Video_Games %>%
filter(Global_Sales > 7.38, Global_Sales < 82.53)
ggplot(Excluding_number1, aes(x=Platform, y=Global_Sales,
fill=Platform))+
geom_boxplot(alpha=0.6)+
labs(x="Platform", y="Total Amount of Copies Sold",
title= "Excluding Number 1")+
theme(legend.position = "none")+
theme(panel.background = element_blank(),
panel.grid.major = element_blank(),
panel.grid.minor = element_blank())+
theme(plot.title = element_text(hjust = 0.5, face="bold"))
To have a better glance of the data, we excluded the number one of the top to work in a better range (2~100).
#Main_Platforms
Main_Platforms <- Excluding_number1 %>%
filter(Platform %in% c("DS","GB","NES","PS2","Wii","X360"))
ggplot(Main_Platforms, aes(x=Platform, y=Global_Sales,
fill=Platform))+
geom_boxplot(alpha=0.6)+
labs(x="Platform", y="Total Amount of Copies Sold",
title= "Global Sales by Platform")+
theme(legend.position = "none")+
theme(panel.background = element_blank(),
panel.grid.major = element_blank(),
panel.grid.minor = element_blank())+
theme(plot.title = element_text(hjust = 0.5, face="bold"))
Again, since over 90% of our top is represented by 6 main platforms, we focused in the most representative ones (“DS”,“GB”,“NES”,“PS2”,“Wii”,“X360”).
#Rating
ggplot(Main_Platforms, aes(x=Rating, y=Global_Sales,
fill=Rating))+
geom_boxplot(alpha=0.6)+
labs(x="Rating", y="Total Amount of Copies Sold",
title= "Global Sales by Rating")+
theme(legend.position = "none")+
theme(panel.background = element_blank(),
panel.grid.major = element_blank(),
panel.grid.minor = element_blank())+
theme(plot.title = element_text(hjust = 0.5, face="bold"))
Classifying by rating, we found out that games from the category “E” (For everyone) are in the first place regarding total amount of copies sold, following “M” (Mature +17y) and “T” (Teen +13y). In the last place is “E10+” with games that are usually for kids with mild violence, if any.
#Genre
ggplot(Main_Platforms, aes(x=Genre, y=Global_Sales,
fill=Genre))+
geom_boxplot(alpha=0.6)+
labs(x="Genre", y="Total Amount of Copies Sold",
title= "Global Sales by Genre")+
theme(legend.position = "none")+
theme(panel.background = element_blank(),
panel.grid.major = element_blank(),
panel.grid.minor = element_blank())+
theme(plot.title = element_text(hjust = 0.5, face="bold"))
Filtering our data by genre, the top category is “Platform” (this is a sub-genre of action video games), so we should consider “Action” category like the winner here. Following in the next places we found out “Racing”, “Sports” and “Puzzle” categories. (Although similar in some aspects, “Role Playing” is different to “Action”)
#pairs_and_correlation_by_Region
pairs(Main_Platforms[,c(3,6:10)])
cor(Main_Platforms[,c(3, 6:10)],use= "complete.obs")## Year_of_Release NA_Sales EU_Sales JP_Sales Other_Sales
## Year_of_Release 1.0000000 -0.50149180 0.1211181 -0.35589108 0.11268879
## NA_Sales -0.5014918 1.00000000 0.2540051 0.21467092 0.07540978
## EU_Sales 0.1211181 0.25400505 1.0000000 0.42476773 0.13167708
## JP_Sales -0.3558911 0.21467092 0.4247677 1.00000000 -0.05647155
## Other_Sales 0.1126888 0.07540978 0.1316771 -0.05647155 1.00000000
## Global_Sales -0.3715746 0.83168126 0.6616454 0.56246632 0.28107735
## Global_Sales
## Year_of_Release -0.3715746
## NA_Sales 0.8316813
## EU_Sales 0.6616454
## JP_Sales 0.5624663
## Other_Sales 0.2810773
## Global_Sales 1.0000000Looking for some patterns or trends in our data regarding sales by region (North America, Europe, Japan and Other regions), we did not find some big correlations between variables. However, it is worth mentioning that we found out that in the last years sales in North America increased more than other regions.
#pairs_and_correlation_by_Score
pairs(Main_Platforms[,c(3,10,11,13)])
cor(Main_Platforms[,c(3,10,11,13)],use= "complete.obs")## Year_of_Release Global_Sales Critic_Score User_Score
## Year_of_Release 1.0000000 -0.10614012 -0.2156673 -0.51529857
## Global_Sales -0.1061401 1.00000000 -0.1731977 0.07777366
## Critic_Score -0.2156673 -0.17319774 1.0000000 0.44337811
## User_Score -0.5152986 0.07777366 0.4433781 1.00000000Similar to above, regarding sales and score correlation we did not find significant results, as expected, while sales increased worldwide through the years, the quantity of “user scores” and “critic scores” grew too.
#Global_Sales_by_Platform
ggplot(Main_Platforms, aes(x=Year, y=Global_Sales))+
labs(x="Year of Release", y="Total Amount of Copies Sold",
title= "Global Sales of All Time by Platform")+
geom_point()+
facet_wrap(~Platform)+
theme(plot.title = element_text(hjust = 0.5, face="bold"))
In the next stage, sorting global sales by platform trough the years, we could confirm that in the “80s” and “90s”, “GB” and “NES” platforms represented almost all sales. In the “00s” most sales belong to “DS”,“PS2” and “Wii”, and “X360” is the winner in the “10s”.
#North_America_Sales_by_Platform
ggplot(Main_Platforms, aes(x=Year, y=NA_Sales))+
labs(x="Year of Release", y="Total Amount of Copies Sold",
title= "North America Sales of All Time by Platform")+
geom_point()+
facet_wrap(~Platform)+
theme(plot.title = element_text(hjust = 0.5, face="bold"))
Since “North America” region is the biggest market worldwide, 50~60% of global sales by platform belong to this region.
#Europe_Sales_by_Platform
ggplot(Main_Platforms, aes(x=Year, y=EU_Sales))+
labs(x="Year of Release", y="Total Amount of Copies Sold",
title= "Europe Sales of All Time by Platform")+
geom_point()+
facet_wrap(~Platform)+
theme(plot.title = element_text(hjust = 0.5, face="bold"))
Comparing “Europe” to “North America” region, we could outline that in the “80s”, “GB” and “NES” sales were more than the double of “Europe” sales. Also, “PS2” sales in the “00s” and “X360” sales in the “10s” were superior in “North America”. However, “DS” sales were almost the same in the “00s”, around 10 million copies sold in both regions.
#Japan_Sales_by_Platform
ggplot(Main_Platforms, aes(x=Year, y=JP_Sales))+
labs(x="Year of Release", y="Total Amount of Copies Sold",
title= "Japan Sales of All Time by Platform")+
geom_point()+
facet_wrap(~Platform)+
theme(plot.title = element_text(hjust = 0.5, face="bold"))
In the “80s” and “90s”, “GB” and “NES” sales in “Japan” were relatively strong (locally), but in the last years overall sales in all platforms are almost immaterial.
#Other_Regions_Sales_by_Platform
ggplot(Main_Platforms, aes(x=Year, y=Other_Sales))+
labs(x="Year of Release", y="Total Amount of Copies Sold",
title= "Other Regions Sales of All Time by Platform")+
geom_point()+
facet_wrap(~Platform)+
theme(plot.title = element_text(hjust = 0.5, face="bold"))
Finally, in “Other Regions” we could find out that only in the “00s”, “PS2” sales made a significant contribution to the “global”.
#Coefficient_of_variation_Global
stdev <- sd(Main_Platforms$Global_Sales, na.rm = TRUE)
stdev## [1] 8.230985avg <- mean(Main_Platforms$Global_Sales, na.rm = TRUE)
avg## [1] 16.36579Coefvar <- (stdev/avg)*100
Coefvar## [1] 50.29384In the last stage, we calculated the standard deviation, mean and coefficient of variation by region and global sales. Since around 25% is an acceptable % that means a low-variance, our “global” coefficient of variation of 50.29% can be considered a medium-variance %. Hence, we can expect sales in coming years can be volatile, by any means stable.
#Coefficient_of_variation_North_America
stdev_NA <- sd(Main_Platforms$NA_Sales, na.rm = TRUE)
stdev_NA## [1] 5.499128avg_NA <- mean(Main_Platforms$NA_Sales, na.rm = TRUE)
avg_NA## [1] 8.288772Coefvar_NA <- (stdev_NA/avg_NA)*100
Coefvar_NA## [1] 66.3443ggplot()+
geom_line(data = Main_Platforms,
mapping = aes(x=Year_of_Release, y=Global_Sales),
color="Black")+
geom_point(data = Main_Platforms,
mapping = aes(x=Year_of_Release, y=Global_Sales),
color="Black")+
geom_line(data = Main_Platforms,
mapping = aes(x=Year_of_Release, y=NA_Sales),
color="Blue")+
geom_point(data = Main_Platforms,
mapping = aes(x=Year_of_Release, y=NA_Sales),
color="Blue")+
labs(x="",y="Total Amount of Copies Sold",
title="Global vs North America Sales")+
theme(plot.title = element_text(hjust = 0.5, face="bold"))
Calculating the coefficient of variation for “North America” region, we got a slightly higher % of 66.34% what means a bigger variance than “global”. Also, we created a graph where we could confirm that most of the time “North America” sales had a big contribution to the “global, except in the”90s” when contribution level decreased.
#Coefficient_of_variation_Europe
stdev_EU <- sd(Main_Platforms$EU_Sales, na.rm = TRUE)
stdev_EU## [1] 2.83954avg_EU <- mean(Main_Platforms$EU_Sales, na.rm = TRUE)
avg_EU## [1] 4.283158Coefvar_EU <- (stdev_EU/avg_EU)*100
Coefvar_EU## [1] 66.29549ggplot()+
geom_line(data = Main_Platforms,
mapping = aes(x=Year_of_Release, y=Global_Sales),
color="Black")+
geom_point(data = Main_Platforms,
mapping = aes(x=Year_of_Release, y=Global_Sales),
color="Black")+
geom_line(data = Main_Platforms,
mapping = aes(x=Year_of_Release, y=EU_Sales),
color="Green")+
geom_point(data = Main_Platforms,
mapping = aes(x=Year_of_Release, y=EU_Sales),
color="Green")+
labs(x="",y="Total Amount of Copies Sold",
title="Global vs Europe Sales")+
theme(plot.title = element_text(hjust = 0.5, face="bold"))
In the case of “Europe”, the coefficient of variation is similar to “North America” region, a 66.30%, a higher variance than “global”. In the graph, we could confirm that “Europe” contribution to the “global” was low, specially in the “80s” and “90s”.
#Coefficient_of_variation_Japan
stdev_JP <- sd(Main_Platforms$JP_Sales, na.rm = TRUE)
stdev_JP## [1] 2.335691avg_JP <- mean(Main_Platforms$JP_Sales, na.rm = TRUE)
avg_JP## [1] 2.350877Coefvar_JP <- (stdev_JP/avg_JP)*100
Coefvar_JP## [1] 99.35404ggplot()+
geom_line(data = Main_Platforms,
mapping = aes(x=Year_of_Release, y=Global_Sales),
color="Black")+
geom_point(data = Main_Platforms,
mapping = aes(x=Year_of_Release, y=Global_Sales),
color="Black")+
geom_line(data = Main_Platforms,
mapping = aes(x=Year_of_Release, y=JP_Sales),
color="Red")+
geom_point(data = Main_Platforms,
mapping = aes(x=Year_of_Release, y=JP_Sales),
color="Red")+
labs(x="",y="Total Amount of Copies Sold",
title="Global vs Japan Sales")+
theme(plot.title = element_text(hjust = 0.5, face="bold"))
In the case of “Japan”, the coefficient of variation is very high, a 99.35%. Regarding contribution to the “global”, most of the time was low.
#Coefficient_of_variation_Other_Regions
stdev_Other <- sd(Main_Platforms$Other_Sales, na.rm = TRUE)
stdev_Other## [1] 1.656475avg_Other <- mean(Main_Platforms$Other_Sales, na.rm = TRUE)
avg_Other## [1] 1.44193Coefvar_Other <- (stdev_Other/avg_Other)*100
Coefvar_Other## [1] 114.879ggplot()+
geom_line(data = Main_Platforms,
mapping = aes(x=Year_of_Release, y=Global_Sales),
color="Black")+
geom_point(data = Main_Platforms,
mapping = aes(x=Year_of_Release, y=Global_Sales),
color="Black")+
geom_line(data = Main_Platforms,
mapping = aes(x=Year_of_Release, y=Other_Sales),
color="Yellow")+
geom_point(data = Main_Platforms,
mapping = aes(x=Year_of_Release, y=Other_Sales),
color="Yellow")+
labs(x="",y="Total Amount of Copies Sold",
title="Global vs Other Regions Sales")+
theme(plot.title = element_text(hjust = 0.5, face="bold"))
Finally, calculating the coefficient of variation for “Other Regions”, we got the highest % of the group, a 114.88%. We also found out that only around 2004 it contributed to the “global” significantly, what confirmed our results above that in the “00s”, “PS2” sales in this region were material.
CONCLUSIONS
After analyzing this dataset, we got some interesting outcomes.
“North America” region is the biggest market and most of the time has had a big level of contribution to the “global”.
While “global” sales have kept increasing through the years, “Japan” and “Other regions” sales have kept a low level of contribution to the “global”.
As expected, in this industry technology and innovation is the most important factor. Every decade has been dominated for “one” or “two” platforms (consoles), making sales of old consoles immaterial to the “global”.
Independently of genre, platform, rating or decade, unexpectedly a video game can become an “outlier” making millions of sales, even if similar games in the past could not get that level of sales.
CONSIDERATIONS
We should be aware that conclusions are based on specific outcomes of this dataset. Since some platforms and best-selling video games are not included here, we can not confirm 100% that these results are accurate.
“Minecraft” with 238 millions of copies sold is not included in this dataset.
“Multi-platform”, “PC” and “Mobile” category of platforms are not included in this dataset.
“Japan” population is smaller than “North America” and “Europe”, so it is logical that contribution level is lower too.