Assignment 8

Research Question

Supercell’s two most popular games are Clash Royale and Clash of Clans. Recently, Clash Royale has been receiving some backlash for transitioning to a pay-to-win structure.

Because of this, I want to see if there is more of a negative view towards Clash Royale compared to Clash of Clans.

Data Source

To obtain this data, I will be scraping product reviews from Metacritic for both Clash Royale and Clash of Clans. This will allow us to collect user reviews, which should contain some emotion and will be adequate for doing sentiment analysis.

Data Collection + Cleaning

First, we will need to install some packages that will allow us to perform these tasks.

Originally, to scrape this data, we needed to create a function that we could use for both Clash Royale and Clash of Clans. Then, we can loop through multiple pages using this function and combine these pages into a single data set. Within each page, we will be scraping three separate vectors: review text, user score(out of ten), and the review date.

Because some of these page reviews can vary, we need to subset each vector to the same length so that each row still represents a single review. After binding all of the pages together, we can filter the data so that it only includes rows with non-missing review text. Finally, we can create a review id for each review.

Once our data is collected and saved to our one drive, we can host it from a static source.

static_clash <- read_csv("https://myxavier-my.sharepoint.com/:x:/g/personal/larkk_xavier_edu/IQCap__E8hNaSI_TKJ3qz1HvASOhkWt6NAwGWzrN_9YSn6Q?download=1")

Text Analysis

Visualization #1

# A tibble: 2 × 6
  game           n_reviews mean_score mean_positivity mean_negativity
  <chr>              <int>      <dbl>           <dbl>           <dbl>
1 Clash Royale         132       2.70            3.02            2.66
2 Clash of Clans       138       7.02            3.59            2.67
# ℹ 1 more variable: positive_per_negative <dbl>

This table shows us the average score, average positive words per review, average negative words per review, and the amount of positive words per negative word. One point to be made is that the average review score for Clash Royale is just 2.7/10, whereas the average score for Clash of Clans is 7.0/10. Both games tend to contain the same amount of negative words, but Clash of Clans has .2 more positive words per negative word than Clash Royale.

Visualization #2

If Clash Royales scores are so low, but the reviews contain the same amount of negativity as Clash of Clans, maybe there are some non-emotional factors that are playing into these low scores, such as Clash Royale being pay to win.

These two charts show maps of the most used words in these reviews, and how often they are used together. As we can see from the word map of Clash Royale, the most linked to word is pay, which would suggest that a lot of people are talking about Clash Royale becoming pay to win. On the other hand, Clash of Clans has nothing mentioned about paying, and reviews focus on it being a fun mobile game.

Visualization #3

This visualization categorizes text from these reviews into emotional groups. As we can see, Clash of Clans reviews contain more words that coincide with emotions such as positive, joyful, and anticipation. On the flip side, Clash Royales reviews contain more words that coincide with emotions such as sadness, disgust, and fear.

From this chart, we can see that reviewers of Clash Royale feel much more negatively about their game of choice than Clash of Clans reviewers.

Visualization #4

This chart looks at the effect a users review score has on their positivity score. From the chart, we can see that as the review score of Clash of Clans goes up, the positivity score also goes up. This suggests that sentiment analysis aligns well with the user score and that better ratings are often accompanied by more positive language.

The same is not true for Clash Royale. For some reason, as the review score for Clash Royale goes up, the level of positivity goes down. Clash Royale has reviews with both positive and negative language at nearly every score level, which implies that text sentiment is more variable and sometimes does not match the numeric rating as closely.

Visualization #5

This chart shows the average positivity score by month over a long period of time. For Clash of Clans, the positivity scores jump up and down quite a bit. There are months with scores as low as 7 and as high as 7. Overall, there isn’t a clear trend and instead there are irregular spikes. This could suggest that players emotions towards Clash of Clans are very volitile and could be a response to new updates. For Clash Royale, the same seems to be true as the positivity scores fluctuate monthly, however, they are are not as volatile as Clash of Clans.

In the past months, the two games have hovered around eachother, right around a positivity score of 0. This tells us that recently, neither game has dominated the other in terms of emotional tone, and instead, both games fluctuate on their own as compared to having a clear long term trend.

Conclusion

Overall, I think it is clear that Clash of Clans is viewed much more favorably than Clash Royale. Clash of Clans has a much higher average user score than Clash Royale(7/10 vs 2.7/10). Reviews of Clash of Clans focused on it being a fun mobile game whereas Clash Royale reviews focused on words like pay, money and game. The emotional bar chart showed us that Clash Royale reviews contain more negative emotions and Clash of Clans reviews tend to contain more positive emotion.

I think it would be reasonable to conclude that Clash Royales decision to implement a pay to win style of gameplay certainly shows in its reviews and has had a large impact on its user review scores.