paulopontesm/titanic

Dataset Card for Titanic Survival Prediction

Dataset Details

Dataset Description

This dataset is a copy of the original Kaggle Titanic dataset made to explore the Hugging Face Datasets feature.

The Titanic Survival Prediction dataset is widely used in machine learning and statistics. It originates from the Titanic: Machine Learning from Disaster competition on Kaggle. The dataset consists of passenger details from the RMS Titanic disaster, including demographic and ticket-related attributes, with the goal of predicting whether a passenger survived.

• —Curated by: Kaggle
• —Funded by: Kaggle
• —Shared by: Kaggle
• —Language(s) (NLP, if applicable): English
• —License: Subject to Competition Rules

Dataset Sources

• —Repository: Kaggle Titanic Competition
• —Paper [optional]: None
• —Demo [optional]: Not applicable

Uses

Direct Use

The dataset is primarily used for:

• —Supervised learning: Predicting survival outcomes based on passenger characteristics.
• —Feature engineering: Extracting new insights from existing features.
• —Data preprocessing techniques: Handling missing values, encoding categorical variables, and normalizing data.
• —Benchmarking machine learning models: Logistic regression, decision trees, random forests, neural networks, etc.

Out-of-Scope Use

This dataset is not meant for:

• —Real-world survival predictions: It is based on a historical dataset and should not be used for real-world survival predictions.
• —Sensitive or personally identifiable information analysis: The dataset does not contain modern personal data but should still be used responsibly.

Dataset Structure

The dataset consists of three CSV files:

1. 1.train.csv (891 entries) – Includes the "Survived" column as labels for training.
2. 2.test.csv (418 entries) – Used for evaluation, with missing "Survived" labels.
3. 3.gender_submission.csv – A sample submission file assuming all female passengers survived.

Data Dictionary

Dataset Creation

Curation Rationale

The dataset was created to help users develop predictive models for classification tasks and serves as an entry-level machine learning dataset.

Source Data

Data Collection and Processing

The dataset originates from historical records of the RMS Titanic disaster and has been structured for machine learning purposes. Some entries contain missing values, particularly in Age and Cabin, requiring imputation or removal.

Who are the source data producers?

The dataset is derived from Titanic passenger records.

Annotations

Annotation process

The dataset is not annotated beyond the Survived label.

Who are the annotators?

The survival labels come from historical records.

Personal and Sensitive Information

The dataset does not contain sensitive or personally identifiable information.

Bias, Risks, and Limitations

The dataset represents historical biases in survival rates:

• —Women and children had a higher chance of survival due to evacuation priorities.
• —First-class passengers had a higher survival rate compared to lower-class passengers.
• —Some data is missing or estimated, particularly age and cabin numbers.

Recommendations

• —Use fairness metrics when training models to assess potential biases.
• —Avoid real-world applications for decision-making, as this is a historical dataset.

Citation

Since this dataset originates from Kaggle, it does not have an official citation. However, you can reference it as follows:

APA: Kaggle. (n.d.). Titanic - Machine Learning from Disaster. Retrieved from https://www.kaggle.com/competitions/titanic/data

BibTeX:

@misc{kaggle_titanic,
  title = {Titanic - Machine Learning from Disaster},
  author = {Kaggle},
  year = {n.d.},
  url = {https://www.kaggle.com/competitions/titanic/data}
}