06. Data exploration

Videogames

• Load packages:

library(dplyr)
library(summarytools)
library(ggplot2)

• Load the data:

Videogames <- read.csv("Video_Games_Sales_as_at_22_Dec_2016.csv")

Contingency tables or crosstab - overview

• Data screening

view(dfSummary(Videogames))

• Crosstab for absolute frequencies - table way:

table(Videogames$Genre, 
      Videogames$Rating)

              
                      AO    E E10+   EC  K-A    M   RP    T
                  2    0    0    0    0    0    0    0    0
  Action       1182    1  416  481    1    0  608    0  681
  Adventure     857    0  162   68    2    0   99    0  115
  Fighting      411    0    8   19    0    0   49    0  362
  Misc          868    0  457  167    5    1   13    0  239
  Platform      319    0  358  144    0    0    3    0   64
  Puzzle        238    0  289   43    0    0    0    0   10
  Racing        377    0  585   96    0    0   18    1  172
  Role-Playing  723    0   84  111    0    0  162    0  420
  Shooter       304    0   48   58    0    0  565    0  348
  Simulation    305    0  326   48    0    0    5    0  190
  Sports        839    0 1188  107    0    0   16    0  198
  Strategy      344    0   70   78    0    2   25    2  162

Bar chart

• Select games without any age restrictions and games with the age restriction for teens, dplyr way:

glimpse(Videogames %>%
          filter(Rating %in% c("E", "T")) %>%
          droplevels(), 
    width = 50)

Observations: 6,952
Variables: 16
$ Name            <fct> Wii Sports, Mario Kar...
$ Platform        <fct> Wii, Wii, Wii, DS, Wi...
$ Year_of_Release <fct> 2006, 2008, 2009, 200...
$ Genre           <fct> Sports, Racing, Sport...
$ Publisher       <fct> Nintendo, Nintendo, N...
$ NA_Sales        <dbl> 41.36, 15.68, 15.61, ...
$ EU_Sales        <dbl> 28.96, 12.76, 10.93, ...
$ JP_Sales        <dbl> 3.77, 3.79, 3.28, 6.5...
$ Other_Sales     <dbl> 8.45, 3.29, 2.95, 2.8...
$ Global_Sales    <dbl> 82.53, 35.52, 32.77, ...
$ Critic_Score    <int> 76, 82, 80, 89, 58, 8...
$ Critic_Count    <int> 51, 73, 73, 65, 41, 8...
$ User_Score      <fct> 8, 8.3, 8, 8.5, 6.6, ...
$ User_Count      <int> 322, 709, 192, 431, 1...
$ Developer       <fct> "Nintendo", "Nintendo...
$ Rating          <fct> E, E, E, E, E, E, E, ...

• Bar chart - absolute frequencies, genres of videogames by ratings

# Data
Videogames_Everyone_Teens <- Videogames %>%
                        filter(Rating %in% c("E", "T")) %>%
                        droplevels() # drop unused levels 
# Plot
ggplot(Videogames_Everyone_Teens, aes(x = Genre, fill = Rating)) + 
  geom_bar(position = "dodge")

• Bar chart - absolute frequencies, genres of videogames by ratings - changing the X-axis’s text angle:

Videogames_Everyone_Teens_Bar_Plot <- Videogames %>%
                              filter(Rating %in% c("E", "T")) %>%
                              droplevels() %>%
  ggplot(aes(x = Genre, fill = Rating)) + 
    geom_bar(position = "dodge") + 
    theme(axis.text.x = element_text(angle = 90))
Videogames_Everyone_Teens_Bar_Plot

Contingency tables or crosstab - relative frequencies

• Crosstab with rating and genres:

GenreRating = table(Videogames_Everyone_Teens$Genre, 
                    Videogames_Everyone_Teens$Rating)
GenreRating

              
                  E    T
  Action        416  681
  Adventure     162  115
  Fighting        8  362
  Misc          457  239
  Platform      358   64
  Puzzle        289   10
  Racing        585  172
  Role-Playing   84  420
  Shooter        48  348
  Simulation    326  190
  Sports       1188  198
  Strategy       70  162

• Relative frequenices - total:

prop.table(GenreRating)

              
                         E           T
  Action       0.059838895 0.097957422
  Adventure    0.023302647 0.016542002
  Fighting     0.001150748 0.052071346
  Misc         0.065736479 0.034378596
  Platform     0.051495972 0.009205984
  Puzzle       0.041570771 0.001438435
  Racing       0.084148446 0.024741082
  Role-Playing 0.012082854 0.060414269
  Shooter      0.006904488 0.050057537
  Simulation   0.046892980 0.027330265
  Sports       0.170886076 0.028481013
  Strategy     0.010069045 0.023302647

• Relative frequencies - rows, rounded by 3 digits:

round(prop.table(GenreRating, 1), 3)

              
                   E     T
  Action       0.379 0.621
  Adventure    0.585 0.415
  Fighting     0.022 0.978
  Misc         0.657 0.343
  Platform     0.848 0.152
  Puzzle       0.967 0.033
  Racing       0.773 0.227
  Role-Playing 0.167 0.833
  Shooter      0.121 0.879
  Simulation   0.632 0.368
  Sports       0.857 0.143
  Strategy     0.302 0.698

• Relative frequencies - columns, rounded by 2 digits:

round(prop.table(GenreRating, 2), 2)

              
                  E    T
  Action       0.10 0.23
  Adventure    0.04 0.04
  Fighting     0.00 0.12
  Misc         0.11 0.08
  Platform     0.09 0.02
  Puzzle       0.07 0.00
  Racing       0.15 0.06
  Role-Playing 0.02 0.14
  Shooter      0.01 0.12
  Simulation   0.08 0.06
  Sports       0.30 0.07
  Strategy     0.02 0.05

Bar chart for relative frequencies

• Proportions of genres by rating:

ggplot(Videogames_Everyone_Teens, aes(x = Genre, fill = Rating)) +
 geom_bar(position = "fill") +
 ylab("proportion")

Bar chart - single variable and a grid

• Add labels to the factor levels:

Videogames_Everyone_Teens$Rating <- factor(Videogames_Everyone_Teens$Rating,
                                      levels = c("E", "T"), 
                                      labels = c("Everyone", "Teen"))

• Bar chart for the number of rating:

ggplot(Videogames_Everyone_Teens, 
       aes(x = Rating)) + 
  geom_bar()

• Bar chart for genres by ratings:

ggplot(Videogames_Everyone_Teens, 
       aes(x = Genre)) +
 geom_bar() +
 facet_wrap(~ Rating) +
 theme(axis.text.x = element_text(angle = 90))

Plots

• Histogram with facets (layers):

ggplot(Videogames_Everyone_Teens, 
       aes(x = Critic_Score)) +
 geom_histogram() +
 facet_wrap(~ Rating)

• Filter games by genres: shooters (action), strategies and RPGs:

Shooter_Strategy_RPG <- filter(Videogames, 
                               Genre %in% c("Shooter", 
                                            "Strategy", 
                                            "Role-Playing"))
glimpse(Shooter_Strategy_RPG, 
        width = 50)

Observations: 3,506
Variables: 16
$ Name            <fct> Pokemon Red/Pokemon B...
$ Platform        <fct> GB, NES, GB, DS, GBA,...
$ Year_of_Release <fct> 1996, 1984, 1999, 200...
$ Genre           <fct> Role-Playing, Shooter...
$ Publisher       <fct> Nintendo, Nintendo, N...
$ NA_Sales        <dbl> 11.27, 26.93, 9.00, 6...
$ EU_Sales        <dbl> 8.89, 0.63, 6.18, 4.4...
$ JP_Sales        <dbl> 10.22, 0.28, 7.20, 6....
$ Other_Sales     <dbl> 1.00, 0.47, 0.71, 1.3...
$ Global_Sales    <dbl> 31.37, 28.31, 23.10, ...
$ Critic_Score    <int> NA, NA, NA, NA, NA, N...
$ Critic_Count    <int> NA, NA, NA, NA, NA, N...
$ User_Score      <fct> , , , , , , 3.4, , , ...
$ User_Count      <int> NA, NA, NA, NA, NA, N...
$ Developer       <fct> "", "", "", "", "", "...
$ Rating          <fct> , , , , , , M, , , M,...

• Box plot:

ggplot(Shooter_Strategy_RPG, 
       aes(x = as.factor(Genre), 
           y = Critic_Score)) +
 geom_boxplot()

• Density plot with overlaying categories:

ggplot(Shooter_Strategy_RPG, 
       aes(x = Critic_Score,
       fill = as.factor(Genre))) +
 geom_density(alpha = .3)

• Histogram for the millions of sold copies videogames in the EU:

Videogames %>%
 ggplot(aes(x = EU_Sales)) +
 geom_histogram(binwidth = 0.01) +
 xlim(c(0, 2)) + 
 ylim(0, 1750) + 
 ggtitle("Millions of sold copies videogames in the EU")

• Histogram for the millions of sold copies of sport videogames in the EU:

Videogames %>%
 filter(Genre == "Sports") %>%
 ggplot(aes(x = EU_Sales)) +
 geom_histogram(binwidth = 0.1) +
 xlim(c(0, 3)) +
 ylim(0, 500) + 
 ggtitle("Histogram for the millions of sold copies of sport videogames in the EU")

• Boxplot for the number of users who gave the user_score**:

Videogames %>%
 ggplot(aes(x = 1, y = User_Count)) +
 geom_boxplot()

• Boxplot for the number of users who gave the user_score* - without outliers*:

Videogames_no_out <- Videogames %>%
  mutate(User_Outliers = 1.5 * IQR(User_Count, na.rm = TRUE)) %>%
  filter(User_Count <= User_Outliers) %>%
 ggplot(aes(x = 1, y = User_Count)) +
 geom_boxplot()
Videogames_no_out

Summaries

• Let’s focus on Ubisoft:

Ubisoft_Reviews <- filter(Videogames, Developer == "Ubisoft") %>%
 ggplot(aes(x = 1, y = User_Count)) +
 geom_boxplot()
Ubisoft_Reviews

• Mean and median of the copies sold in the North America by Ubisoft, by genres:

Videogames %>%
 filter(Developer == "Ubisoft") %>%
 group_by(Genre) %>%
 summarize(round(mean(NA_Sales), 3),
 median(NA_Sales))

• Box plot of the number of copies sold in the North America by Ubisoft, by genre:

Ubisoft <- Videogames %>%
            filter(Developer == "Ubisoft")
Ubisoft %>%
 ggplot(aes(x = Genre, y = NA_Sales)) +
 geom_boxplot()