• Anime ratings can vary widely between shows and it is based on multiple factors
• This project focuses on what could influence these ratings

  

• Dataset: MyAnimeList Anime Dataset
• MyAnimeList is a website that is used to track, categorize, rate, and discuss anime
  
• It contains information on anime ratings, popularity, and other characteristics
• The dataset was found on Kaggle
• Key variables used in this analysis:
  • score (rating)
  • episodes (number of episodes)
  • members (popularity)
  • type (TV, Movie, OVA)

library(tidyverse)
library(knitr)


data = read_csv("anime.csv")

#cleaning the dataset
data = data %>%
  select(title, score, episodes, members, type, sfw) %>%
  filter(sfw) %>%
  select(-sfw) %>%
  filter(type != "music") %>%
  filter(type != "special") %>%
  drop_na()

• The “Members” column shows how many people have it on their list
  • A member could put an anime in one of the following categories on their list: Watching, Completed, On Hold, Dropped, and Plan to Watch

• ONA: Original Net Animation (straight to streaming services)
• OVA: Original Video Animation (it could extra episodes that are not from the source material, or its not based on existing media)

• OVAs and ONAs have a slightly lower average rating
  • For OVAs, it could be attributed to the fact that it is not adapted from canon material and won’t be as good as the original story.
  • For ONAs, it could be due to the lack of marketing. Fewer viewers could lead to fewer ratings.
    
• OVAs and ONAs also have outliers at the top and bottom which shows that there are some exceptional OVAs and ONAs out there
• Movies and TV shows having a higher rating makes sense as they have higher budgets and more backing from different companies.
• It makes sense that the highest rated media are TV shows since they can run for much longer and have more character development.

• The data is approximately normal, skewing slightly positive
• Since the peak is above 5, it could indicate that users are more generous when it comes to a media they like
• The reason for a staggeringly low amount of ratings below 5 could be due to the loss of interest
  • If an anime is not that good, someone could just drop it entirely and move on to something different
• The lack of ratings below 3.5 could suggest selection bias
  • Viewers are much less likely to watch low rated anime and it could lead to the under representation of scores

• The graph shows that having more episodes does not lead to a higher score
• There are plenty of anime with episode counts in the hundreds that have an average score
• There are some outliers which have more than 1000 episodes, but these are very long running anime from franchises such as Pokemon, Doraemon (a popular kids show in Asia), etc.
• These are mostly kids shows which have been airing for decades
• Overall, this graph tells us that the people behind the anime matter just as much as the number of episodes
• The right team can make a ground breaking anime whether it is only 12 episodes long, or even 200+

#Pearson correlation between score and members
cor_test <- cor.test(data$score, data$members, method = "pearson")

cor_test

## 
##  Pearson's product-moment correlation
## 
## data:  data$score and data$members
## t = 46.162, df = 10865, p-value < 2.2e-16
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  0.3890906 0.4205304
## sample estimates:
##       cor 
## 0.4049301

• The result of 0.405 shows a moderate positive correlation
• Significance: p-value < 0.001 shows that the correlation is statistically significant
• Based on the confidence interval, I am confident that the true correlation is positive
• Interpretation: More popular anime tend to have higher ratings, but the relationship is moderate, meaning that other factors also affect scores

Dataset: https://www.kaggle.com/datasets/andreuvallhernndez/myanimelist

Tools Used: R, ggplot2, plotly, dplyer