Video Games Sales and Ratings

Introduction

We are looking a dataset that consists of over 10,000 video games as of January 2017. This data heavily focus on video game ratings from Metacritic and video game sales from VGChartz.

As mentioned in the site, the variables in this dataset include:

• Name
• Platform
• YearofRelease
• Genre
• Publisher
• NA_Sales (North America sales in millions)
• EU_Sales (Europe sales in millions)
• JP_Sales (Japan sales in millions)
• Other_Sales (Sales in rest of the world in millions)
• Global_Sales (Global sales in millions)
• Critic_score (Aggregate score compiled by Metacritic staff)
• Critic_count (Number of Critics)
• User_score (Score by Metacritic’s subscribers)
• User_count (Number of Users who gave user scores)
• Rating

The research questions that we aim to answer are:

• How do video game publishers feel about the different video game genres that they publish?
• Is it important for game publishers to care about critic ratings for sales?
• What is causing Japan’s abnormal video games sales distribution relative to Europe and North America?

Research Question 1: How do video game publishers feel about the different video game genres that they publish?

The research question that we are aiming to answer here is whether publishers in more general and specific contexts prefer to create games of specific video game genres. We will be looking into the distribution of the Genre variable and the relationship between Genre and Publisher.

Is there an equal distribution of different genres of video games that are published?

## 
##  Chi-squared test for given probabilities
## 
## data:  table(videogames$Genre)
## X-squared = 5129, df = 11, p-value < 2.2e-16

This pie chart shows how the different genres are distributed among the different games.

The null hypothesis is that the video games of different genres are of equal proportions. With a p-value of approximately 0, we can reject the null hypothesis and say that the proportions of video games with different genres are not all the same. In other words, generally speaking, by looking at the pie chart, the games of publishers are generally of great variety though there are still some types of games that are typically published more often over other genres. From the pie chart, action games are very high proportionally, followed by sports games. Meanwhile, puzzle, strategy, and fighting games are not as common.

Do the top 10 publishers that publish the most games correlate to different genres?

Given that there were so many publishers and some publishers specialize in specific genres, we have decided to look at the top 10 publishers that make the most games to see if the big video game publishers try to create a variety of equal proportions in terms of video game genres.

## 
##  Pearson's Chi-squared test
## 
## data:  top_10_publishers_v_genre
## X-squared = 3034.7, df = 99, p-value < 2.2e-16

This mosaic plot is between the video game genre and the top 10 video game publishers by number of games they publish.

The null hypothesis suggests that the big video game publishers and the game genres are independent of each other. With a p-value of approximately 0 and the multiple chunks of red and blue in the mosaic plot where there were unusually high and unusually low proportions in certain combinations of video game publisher and video game genre, we can reject the null hypothesis and say that big publishers and video game genres are not independent of each other.

This may be because that certain companies are better at certain genres than others and would rather publish games that they can specialize at. This also demonstrates potentially how different companies have different target audiences. For example, Activision seems to create an unusually high number of action and shooting games, as well as an unusually low number of adventure, role-playing, simulation, sports and strategy games. Meanwhile, Nintendo seems to create an unusually high number of puzzle, role-playing, platform, and miscellaneous video games, as well as an unusually low number of action, fighting, shooter, and sports video games.

Research Question 2: Is it important for game publishers to care about ratings for sales?

Here, we wanted to explore the relationship between game ratings and how well they sold. Because the market tends to be recommendation-based, we wanted to examine how much attention companies should be paying to ratings. Additionally, we wanted to see if there was a discrepancy between two major types of ratings: user ratings, and critic ratings. Accordingly, we examined rating variables (User_score, Critic_score), sales variables (Global_Sales, NA_Sales, EU_Sales, JP_Sales, Other_Sales), and attribute variables (Genre).

Are user and critic ratings positively correlated with each other?

Here is a critic score vs user score contour plots facetted by genre that show the points that each represent a video game and demonstrate the general density of the video games based on the ratings. Overall, most of the genres have one peak in their corresponding contour plot at around the 70-80 mark for the critic scores and around 7.5-8.0 mark for the user scores, and the contour lines are generally very similar throughout the genres. The only contour plot without in the Simulation genre, there is another peak at around 62 for critic score and 5.0 for user score.

Overall, for a lot of these genres, there really isn’t an obvious bias from the Metacritic critics and users as most of the densities are somewhat similar for each genre by looking at these graphs. In addition, it seems like that the games that are typically rated higher by Metacritic users are also highly rated by Metacritic critics. Therefore, these user scores and critic scores are generally positively correlated with each other.

How do critic ratings correlate with sales across different genres?

When comparing across genres, there are much more pronounced differences in the sales for highly rated games. For instance, we can see that action and racing games show extremely large spikes in sales across all regions, while adventure, puzzle, simulation, and strategy all show pretty small spikes in sales across ratings. This perhaps suggests that people’s interest in some game genres is independent of critic ratings. Interestingly, the miscellaneous showed an approximately linear trend in sales; as ratings increased, so did the sales.

Are game titles more closely related in certain regions than others?

## [1] 4000

Next, we wanted to find out clustering patters with the main video game selling regions: North America, Japan, and Europe. For this, in order to take into account all quantitative variables, we used MDS.

From MDS Analysis of each of the main video game sales areas (other areas omitted as results would be less valuable valuable), we found that there is only one cluster for each area’s sales. These clusters also occurred in the same area of MDS plot. This indicates that the sales in each area are correlated similarly by other quantitative values such has critic score/count and user score/count. However, very interestingly, North America and Europe’s cluster shape almost identical in shape and very dense, whereas Japan’s MDS plot had a slightly larger and more wide- spread cluster.

This lines up well with previous results. However, the fact that these clusters are so similar in shape is very unlikely. A possible reason for this is the nature of the dataset itself. There is only one rating count/value for users and critics which is spread across all regions in the dataset uniformly. However in reality, because of differences in media outlets by region, a game rated well by Metacritic, an American Firm, might not have been exactly as well rated by a media outlet in Japan. Note: this does not discount other correlations found as it is unlikely that a game rated very well in one region would be extremely hated in another. However, it does critique the exact accuracy these ratings have, likely causing the more similar shape.

Research Question 3: What is causing Japan’s abnormal video games sales distribution relative to Europe and North America?

Given the different distribution of Japan’s sales in comparison to Europe and North America, we aim to observe potential variables that suggest this abnormality. This includes looking at relationships between the region sales and platform, as well as looking at Nintendo specific data.

How do the different regional sales correlate to video game platforms?

Platforms are the method by which people play games. However, rather than just look at the number of games released per platform in each area, we sought to find out which platforms have the highest sale for games overall as this would be the most beneficial information for video game production firms.

We can see that North America has more overall sales than Europe and Japan. One notable aspect of the data this distribution highlights is the lack of Microsoft Platform use in Japan. The Xbox360 and Xbox One, which are very popular console choices in North America and still notably used in Europe has very few sales in Japan. In contrast, despite the fact that North America constantly outsells for large consoles, for Nintendo platforms, Japan’s sales are notably high. Finally, on the global scale, we can see that Sony consistently outperforms Microsoft for competing platforms (i.e. Xbox360 vs. PS3, PS4 vs Xbox One, etc.).

Does Nintendo as a video game publisher focus their games towards Japan audiences?

This graph on the left shows that the proportion of sales for Nintendo Games follows similar trends as other regions in terms of overall sales over time. The graph on the right shows how the proportion of sales from Japan is very comparable to those of other regions despite the fact that the bar chart shown previously across all publishers showed Japan consistently being outsold by Europe and North America. This further helps show how successful Nintendo is in Japan, especially relative to how they perform globally.

Conclusion

Overall, publishers seem to favor certain genres over others in order to sell well to other users and receive positive feedback from those users as well. In addition, Metacritic critics and users generally seem to share similar thoughts on the games as well. With specifically Nintendo, they seem to have a very specific audience and create games that aim to meet those needs.

In terms of potential future work, it would be interesting to look at more specific publishers and potentially sales of other different regions. Given that there are so many potential publishers in different regions, it would be interesting to see the market different publishers are focusing on. In addition, VGChartz do not have all of the sales data for all of the video games, so it would be interesting to see the missing gaps, as well as potentially not as high selling games. We can look into other publishers and see whether other potential factors can shift the position of the “biggest publisher” globally.

It would also be worth it to look at different rating systems. Given that our analysis was based on ratings from one website, our analysis is highly prone to bias from the people posting on the site, or even just the site itself. Future work could examine different sources of ratings, or even different systems of rating. One avenue we could explore is the correlation between different rating systems. Dimension reduction could prove useful in that respect, especially in looking at how the distance between points might change as more rating systems are added. Another avenue we could explore is the application of these ratings to different regions. It’s worth exploring how correlated rating systems are with certain regions, especially in looking at if different rating systems are better predictors of sales in their own respective regions.