ecopus/pokemon_cards

Dataset Card for Dataset Name

This dataset contains information about real-world Pokémon cards, including both an original set of 34 cards and an augmented set of 300 synthetic cards. It includes features such as card name, year, card set, artwork style, condition, market value, and whether the card is considered a collector's item.

Dataset Details

Dataset Description

This dataset is a collection of real-world Pokémon card data. The original portion consists of 34 real cards with various attributes. The augmented portion was generated synthetically to expand the dataset size to 300 entries by varying existing features and adding a unique identifier to each card. The goal is to provide a larger dataset for potential analysis or machine learning tasks which work to predict a given card's collectors item status.

• Curated by: Emily Copus
• Shared by: ecopus (Hugging Face Hub)
• Language(s) (NLP): English
• License: apache-2.0

Dataset Sources [optional]

• Repository: https://huggingface.co/datasets/ecopus/pokemon_cards

Uses

Direct Use

This dataset can be used for:

• Exploratory data analysis of Pokémon card attributes.
• Training machine learning models to correlate card attributes to identify a card's collector's item status.
• Practicing data manipulation and augmentation techniques.

Out-of-Scope Use

This dataset is not suitable for:

• Making financial decisions based on the market values, as these are based on a small sample and synthetic augmentation.
• Representing a comprehensive and accurate historical record of all Pokémon cards.
• Uses requiring real-time market data (market values are current as of 9/14/2025).

Dataset Structure

The dataset is split into two configurations: 'original' and 'augmented'. Both configurations have the following features:

• Card (string): The name of the Pokémon card. Augmented cards are a random choice of card, and have a random number appended.
• Year (int): The year the card was released or assigned. Augmented years are within a 4 year range of the original year.
• Card Set (string): The set the card belongs to. Augmented cards use existing sets from the original data.
• Artwork Style (string): The style of the card's artwork (e.g., Standard, Full Art, Holo).
• Condition (string): The condition of the card (e.g., Mint, Near Mint).
• Set Number Eq (float): A calculated value related to the set number (which number card it is wrt the full set). For example, if a card is 118 out of 220 total cards for that set, its set number eq is 0.536. Set number eq values greater than one correspond to "Secret Rare cards", thus indicating desirability.
• Market Value (float): The estimated market value of the card, based upon real time values from tcgplayer.com. Augmented values are based on the original with a +/- 0.2 scale factor.
• Collector's Item (string): Indicates if the card is considered a collector's item (Yes/No), based on artwork style, market value, and condition for augmented data. The 'original' split contains 34 examples, and the 'augmented' split contains 385 examples.

Dataset Creation

Curation Rationale

This dataset was curated as a basic learning tool for implementing ML tools with tabular datasets. The simplicity of this dataset allows for easy implementation into basic ML classification algorithms, pefect for a first time user.

Source Data

The source data for this dataset is 34 real-world pokemon cards and their attrubutes.

Data Collection and Processing

The Pokemon cards were chosen randomly from the curator's personal collection, with roughly 50% being collectors items and 50% not. The cards chosen span from 1995-2025 across a range of 22 sets.

Who are the source data producers?

The curator was the producer of this dataset.

Annotation process

Annotations were created based off the physical features printed on the cards, along with the target market value from tcgplayer.com. It is important to note that the collector's item status was based off intimate knowledge of the curator's personal collection - these cards have not been professionally graded. For the augmented dataset, a conditional was used to assign collector's item status to the augmented data. Ultimately, if a card is not heavily played, not standard art, and have a current market value greater than $15, is is assigned a collector's item. [More Information Needed]

Who are the annotators?

The curator is the annotator for this dataset.

Personal and Sensitive Information

This dataset contains no personal nor sensitive information.

Bias, Risks, and Limitations

The original dataset is very small (only 34 cards) and only contains cards from 22 sets, which limits the diversity and representativeness of the data. The augmented dataset, while larger (300 cards), is synthetically generated based on the small original set, meaning it may not fully capture the complexities and nuances of real-world Pokémon card data (many of these likely do not respresent real world cards). The augmentation process involves some randomness and simple rules (like the one for determining 'Collector's Item'), which may not reflect actual market dynamics or collector behavior accurately. Sociotechnically, the market values are estimations and are subject to the limitations of the small original sample and the augmentation process, so relying on them for actual financial decisions is not advisable. The dataset also doesn't account for regional variations in card releases, market values, or collector preferences, which are significant factors in the real world of Pokémon cards.

Recommendations

Users should be aware of the risks, biases and limitations of the dataset before use (see above). Relying on models making predictions from this dataset for actual financial decisions is not advisable. Simply, this tool should be utilized to better make connections between a card's features and its status as a collector's item.

Dataset Card Authors [optional]

Emily Copus

Dataset Card Contact

[email protected]