Mikami063/User-Animelist-Dataset

---
preview_file_name: ratings.csv
license: cc-by-4.0
pretty_name: User Animelist Dataset
size_categories:
- 1M<n<10M
modalities:
- tabular
task_categories:
- tabular-regression
- tabular-classification
language:
- en
tags:
- Anime
- Recommendation
- Tabular
- Dataset
- Recommender
- MovieLens
---

# About Dataset

**This dataset consists of user ratings for anime titles. Each user in the dataset has provided at least 5 ratings,
ensuring a minimum level of engagement. The dataset includes detailed information about both users and anime, making it
suitable for tasks such as recommendation systems, user behavior analysis, and genre-based filtering.
Dataset is freshly-created so it cover newer animes. Data is published in MovieLens format except timestamp data so
this dataset is easy to use with GitHub that trains with MovieLens dataset**

## Dataset Statistics

- Number of Users: 1,774,522
- Number of Animes: 20,237
- Total Ratings: 148,170,496
- Sparsity/Density: 0.0041
- Average Ratings per User: ~83.50
- Average Ratings per Anime: ~7,321.76
- Rating Range: 0.1 to 10.0
- Mean Rating: 7.64
- Standard Deviation of Ratings: 1.89

## Anime Metadata

### Each anime entry includes:

- Title
- Year of release
- Episode Count
- Type (e.g., TV, Movie, OVA)
- Score (aggregated or average rating)
- Image URL (for visual reference)
- MyAnimeList URL
- Genres Detailed Genres

## Usage

```bash
file_path = 'ratings.npy'

# ratings_array shape: (n_ratings, 3) - columns: [user_id, anime_id, rating]
ratings_array = np.load(file_path)

# Create DataFrame from numpy array
df = pd.DataFrame(ratings_array, columns=['user_id', 'anime_id', 'rating'])
```

## Links

Dataset GitHub repo: https://github.com/MRamazan/User-Animelist-Dataset

Dataset Kaggle link: https://www.kaggle.com/datasets/tavuksuzdurum/user-animelist-dataset

BERT Anime Recommender GitHub repo: https://github.com/MRamazan/AnimeRecBERT