Synthetic Health Insurance Claims For Fraud Detection

Synthetic health insurance claims data containing 1 million rows with labeled fraudulent activity. The dataset is designed for fraud detection modeling and originates from Kaggle. The creation date and author are unknown.

Use Cases

• Train binary classification models to predict the 'fraud' label from claim amount and provider details.
• Analyze patterns in claim submission frequency and patient demographics to identify anomalous behavior.
• Build anomaly detection systems using features like procedure codes and billed amounts.
• Test feature engineering techniques on synthetic claim attributes such as diagnosis codes and service dates.

Strengths

• 1 million synthetic rows provide a substantial volume for model training.
• Includes an explicit fraud label for supervised learning tasks.

Limitations

• Data is synthetic, not real-world, limiting generalizability to actual fraud patterns.
• Unknown column structure and feature completeness may require assumptions for modeling.
• Lack of provenance details prevents assessment of synthetic generation quality.

Provenance

Source

Kaggle

Collection Method

Synthetically generated for fraud detection tasks.

Time Range

null

Freshness

null

Geography

null