Introduction

Online games have become a popular form of entertainment, reaching millions of players. Games are dynamic environments, in which the players interact with the game and with other players all over the world. In this sense, games present rich data due to its digital nature. Thus, it is a promising environment to study and apply Artificial Intelligence and Data Mining techniques.

Context

In this Kaggle Dataset, we provide over 115 thousand games maps created on Super Mario Maker with over 880 thousand players which performed over 7 millions of interactions on these maps. By interactions, this means that a player can: (1) create a game map; (2) play a map created by other players; if a player completes the challenge of the game map, he/she (3) "cleared" the map; also can be the (4) first clear; beat the (5) time record of a map; (6) at any time, the player can "like" a game map. Note, this dataset present temporal changes over time for each game map by a period of three months.

The data was extracted from supermariomakerbookmark.nintendo.net, the game website. Now it is publicly to everyone play, explore and research. This dataset serves as a good base for learning models, including, but not limited to, Player Modeling (e.g., player experience), Data Mining (e.g., prediction, and find patterns), and Social Network Analysis (e.g., community detection, link prediction, ranking).

Dataset

This dataset is split into seven files:

• courses.csv: game maps data.
• course-meta.csv: temporal changes on game maps.
• players.csv: players' data.
• plays.csv: plays over time.
• clears.csv: clears over time.
• likes.csv: likes over time.
• records.csv: records over time.

Data Description

The figure illustrates a schema with non-normalized tables to store the SMMnet into a Relational Database Management System (RDBMS). Basically, it is composed of seven tables, each one for one CSV file, that include the maps, players, and the changes over time. Note, there are two Primary Keys (PK) on these tables, i.e., courses.id and players.id for be linked by the Foreign Keys (FK) of other tables to reference and associate with them.

Inspiration

1. Detecting game influencers (e.g., twitch and YouTube users). Work.
2. Predict the popularity of a game over time.
3. Identify popular games characteristics.
4. Player Modeling (e.g., player activity)
5. Social Network Analysis (e.g., community detection, link prediction)
6. Your creativity!

Acknowledgements

• Nintendo Inc., Kyoto, Japan by created this amazing game.
• Photo by Cláudio Luiz Castro on Unsplash.

Citation

SMMNet is available for researchers and data scientists under the Creative Commons BY license. In case of publication and/or public use, as well as any dataset derived from it, one should acknowledge its creators by citing us. Bibtex.