Which is the best Japanese anime studio?
Visualizing anime studios from different perspectives
by Airy Liu and Bhavesh Bellara
Nowadays, anime has become increasingly more mainstream and its market continues to expand all over the world. In 2017, the anime market set a new sales record of $19.8 billion owing greatly to overseas demand (Jozuka, 2019). This fast-growing industry has triggered people’s interest in studying it, and we are also part of them. Based on our online research, some anime fans carried out analysis by themselves out of a passion for anime and the curiosity of the industry. Their works include creating anime rankings, analyzing the changing trend of preferred anime genres, and demographic characteristics of anime audiences (Bilgin, 2019; Rafiq, 2019; Overlytic, 2019). Beyond that, the Association of Japanese Animations has been publishing annual reports about the Japanese Animation Industry, which has become one of the main sources of online analysis about the anime industry (AJA, 2019).
Although these works are insightful, none of them specifically looked at anime studios. There have been some anime studio rankings on the internet while they are only based on subjective online voting (Teffen, 2017; Lindwasser, 2019). Therefore, we are inspired to fill the gap by objectively analyzing anime studio. To be more specific, we will be evaluating: 1) Which anime studio is the most successful studio; 2) What are the characteristics of successful studios?
Data source
We were able to get an exhaustive dataset on anime from Kaggle.com (MyAnimeList Dataset, 2018). This dataset was almost 2Gb in size and contained 3 sub-datasets: ‘Anime List, ‘User List’, ‘UserAnime List’. There was also filtered data and cleaned data versions of these datasets available. Overall, the dataset captures data from 302,675 unique users and 14,478 unique animes.
We mainly made use of the AnimeList dataset, which contains 31 columns such as anime name, anime ID, studio, genre, rating, favorite (how many times an anime was added to a user’s favorites list), popularity (how many people have watched the anime), etc.
Data preparation
Since the original dataset is based on anime, and one anime can be produced by multiple studios that have been stored in one cell, our first step was to split different studios into multiple columns. After that, we used the pivot option to have separate these multiple studios into individual rows. We also split columns that contain multiple genres using the same technique.
Then, in order to comprehensively analyze anime studios, we created a formula that covers different aspects: Overall studio score = popularity score+loyal audience score+ anime quality score+ anime quantity score. For each anime studio, a higher score means a higher rank. To further elaborate on the four criteria: 1) Popularity score (the sum of audiences of all animes the studio produced divided by the number of anime studio produced) to see on average how popular the studio’s animes are; 2) Loyal audience score (the total number of times animes being added to one’s favorite list divided by the sum of audiences of the animes produced by the studio) to see how successfully can the studio’s animes transform a normal audience to a loyal fan; 3) Quality score (the average rating of all animes produced by the studio) to see the average quality of the studio’s work; 4) Quantity score (the total number of animes produced by the studio). Before adding up the four scores, we also standardized them so that they are on the same scale.
So, who is the winner?
After calculating the overall score for all studios, we got our winner: Toei Animation. It has an overall score of 441.3, which is 68 higher than the second studio.
We have also highlighted the top 20 studios that are going to be analyzed in detail in the following paragraphs.
How studios perform based on different criteria we set
We are interested in visualizing not only the overall scores of each studio but also their individual scores. Therefore, we created four scatterplots showing the overall score as well as other criteria at the same time. Each dot refers to one studio (see Figure 2). We also applied colors to show the overall scores of studios, so it is easier to identify patterns: the greener a dot is the higher it is in our ranking (the higher overall score it gets), while the redder it is the lower it is in the ranking.
1. The current anime industry is dominated by mega-studios
It can be seen that there is a huge gap between the top studios and “the normal majorities”. Very few studios are green and the majority of them are in orange or red, which means they receive low scores in our ranking. This result reflects the current anime industry structure, which is dominated by mega-studios while is hard for small studios to survive due to budgetary constraints (Margolis, 2019).
2. Popular studios have different strategies. Some of them aim at the mass market, while others target niche audience groups.
As for the top studios, all of them receive decent quality scores (see Figure 3). However, these studios perform very differently in terms of how popular they are, how many loyal audiences they have got for each produced anime, and the amount of produced anime.
For example, the top one studio: Toei animation has received a high quantity score, which means it has produced a large number of animes (see Figure 4). As a result, the overall popularity of Toei animation is also high. Furthermore, the loyal audience score is also competitive. It means after watching Toei animation’s work, a lot of audiences added it to their favorite. Nevertheless, the average popularity score is relatively low, which shows that not all the anime produced by Toei animation are popular with the general public. However, the situation for Kyoto Animation is totally different. Kyoto Animation ranked 8 in our ranking. Similar to Toei animation, it has a high-quality score, which is even a little bit higher than Toei animation. While Kyoto Animation produces much fewer animes than Toei animation, thus generally speaking it is not that popular among anime audiences. Nonetheless, its average popularity score is much higher, which reflects that even Kyoto Animation is not that “productive”, every anime they produce are high-quality and most of them would become popular among the general public.
How studios collaborate with each other
The dataset also shows a lot of collaboration amongst various studios to produce anime. As Figure 5 shows, we have visualized all the studios in a single graph and highlighted connections between them with the network graph. The size of the node signifies the number of total collaborations the studio has done with other studios. The bigger the size, the higher the number of collaboration. Eg: Madhouse likes to collaborate more when producing animes. While the color of the node signifies the rank of the studio in our calculated ranking. The darker the red shade, the better the rank of the studio in our ranking list. Furthermore, the thickness on the line connecting two studios depends on the number of unique animes these two studios have co-produced. A thicker line means higher collaboration instances between the two studios.
Not all top studios are active collaborators, and they sometimes work with small studios as well.
Unexpectedly, although top studios that ranked high tend to be main collaborators, their popularity among other studios is not always proportional to their overall scores. Furthermore, top studios do not necessarily collaborate only with top studios. For instance, despite being the top one studio, Toei animation collaborates less frequently than studios such as Madhouse, Production I.G., and Sunrise (see Figure 6). The relationship between Toei animation and its collaborators is also weak, as the lines are relatively thin which means they did not collaborate on a lot of animes.
What are the popular genres among all the studios: Action, adventure, and Fantasy
With a wide variety of anime genres present in our dataset, we were interested in finding which was the most popular genre. We did this by creating a word cloud of all the anime genres present in our dataset (see Figure 7). After doing so, we were able to identify that ‘Comedy’ is the most popular genre closely followed by ‘Action’, ‘Adventure’ & ‘Fantasy’ genres. Beyond that, ‘Romance’, ‘Sci-Fi’, ‘Drama’, ‘Slice of Life’, and ‘Kids’ are also main genres.
Analysis of the top 20 studios
Next, we decided to focus only on the top 20 studios, to see if they share any similar strategies that make them successful, or if they have their own unique tactics.
1. Top 3 genres among the top 20 studios are Comedy, Action, and Adventure.
As Figure 8 shows, we decided to produce a column chart to show the genre composition of the top 20 studios to learn more about their strategies. To make the visualization more readable, only the top 10 genres are shown for each studio, while genres out of the top 10 are grouped under ‘Other’. Studios have also been sorted based on their ranks.
It can be seen that most top studios have a very diverse genre composition. They are not limiting themselves to several anime genres. The genre popularity here aligns with one of all the anime studios, in which ‘Comedy’ ranked first followed by ‘Action’ and ‘Adventure’. While the genre ‘Magic’ is more popular among top studios than it is among all the anime studios.
2. Top 20 studios produce mostly PG-13 animes.
Another data field that caught our attention was the ‘Ratings’ column. To find which was the most rating category where the majority of anime from our top 20 studios belonged, we created a treemap (see Figure 9). Doing so helped us identify that the ‘PG-13’ rating category is the most popular rating category with accounting for more than 50% of all the anime produced by the top 20 studios.
3. Top 20 studios produce less than 50 episodes for over half of their animes.
While working on the Top 20 studios, we were amazed to find out the number of episodes these top studios produced for each of their animes. To identify if this was a recurring trend or just an anomaly, we created a box plot where all the animes from these top 20 studios were plotted on a single graph and arranged on the basis of the number of episodes each of them had (see Figure 10).
This visualization highlighted that studios first like to produce less than 50 episodes for one anime and then if these 50 episodes create hype for that show, more episodes are produced. This can be explained based on the fact that 19 out of 20 top studios produced less than 50 episodes for more than 50% of their animes. And then there are some extremely popular anime like Dragon Ball Z, Naruto & Bleach with a very high number of episodes compared to the animes produced by the same studio.
Limitation and Future work
While we were able to unearth some interesting insights from our visualizations, we would still like to acknowledge certain limitations that we faced while working on this project.
- Since our dataset is extracted from MyAnimeList, it is limited to the users of this website, the majority of whom are from the United States, the United Kingdom, Canada, Russia, and Brazil. However, other countries such as Japan and China also have huge consumer groups of anime. While these audiences tend to use different websites to view or review animes, thus are not included in the dataset. Therefore, if we could supplement the dataset with other different datasets, we would be able to get a more comprehensive analysis.
- More facts could be considered in our formula. For instance, we could also compare the studios’ revenue, which would provide us with a business perspective when analyzing the studios. During this project, we tried to gather such data while due to time and source limitations, we were not able to find complete data for it.
Link to the visualization demo
Feel free to interact with our visualization files via these links. While due to the nature of Gephi, we were not able to share the network graph online. If you are interested, feel free to message and ask for the file. Enjoy!
References
AJA (2019). Anime Industry Data. Retrieved December 10, 2019, from https://aja.gr.jp/english/japan-anime-data
Bilgin, F. (2019). Story of Anime. Retrieved December 10, 2019, from https://www.kaggle.com/fatihbilgin/story-of-anime
Jozuka, E. (2019, July 29). How anime shaped Japan’s global identity. Retrieved December 10, 2019,2019, from https://www.cnn.com/style/article/japan-anime-global-identity-hnk-intl/index.html
Lindwasser, A. (2019). The 15 Greatest Anime Studios of All Time, Ranked. Retrieved December 10, 2019, from https://www.ranker.com/list/best-anime-studios-of-all-time/anna-lindwasser
Margolis, E. (2019). The dark side of Japan’s anime industry. Retrieved December 10, 2019, from https://www.vox.com/culture/2019/7/2/20677237/anime-industry-japan-artists-pay-labor-abuse-neon-genesis-evangelion-netflix
MyAnimeList Dataset. (2018). Retrieved December 10, 2019, from
https://www.kaggle.com/azathoth42/myanimelist.
Overlytic. (2019). Anime: The good, the bad and the popular. Retrieved December 10, 2019, from https://airelevant.netlify.com/post/popular_anime/
Rafiq, H. (2019). Analyzing Anime data in R. Retrieved December 10, 2019, from https://towardsdatascience.com/analyzing-anime-data-in-r-8d2c2730de8c
Teffen. (2017, Aug 27). The 10 Most Loved Anime Studios Ranked by Japanese Fans. Retrieved December 10, 2019, from https://goboiano.com/the-10-most-loved-anime-studios-ranked-by-japanese-fans/
