This project demonstrates how to analyze insurance claims data by applying a complete actuarial data science workflow using Python and Tableau. It showcases data cleaning, feature engineering, fraud detection, and stakeholder-ready visualizations.

The goal: Identify potential fraudulent activities, assess customer risk profiles, and deliver insights to stakeholders through a clear visual dashboard.

The goal of this project was to explore the feasibility of using basic feature engineering and rule-based risk scoring to flag potentially fraudulent insurance claims in a simplified dataset.

The focus was on:

• Cleaning and merging claims and customer data
• Creating basic risk flags using thresholds
• Testing with a synthetic fraudulent claim
• Building a proof-of-concept Tableau dashboard for stakeholder presentation

• Loaded raw datasets: Insurance_Claims_Data.csv and Insurance_Customers_Data.csv
• Inspected for missing values, data types, and duplicates
• Standardized column names and merged datasets on customer_id
• Saved the result to Cleaned_Merged_Insurance_Data.csv

• Feature engineered new metrics:
  • claim_to_premium_ratio
  • is_high_risk (binary flag)
  • processing_efficiency
• Exported as Feature_Engineered_Insurance_Data_with_ClaimID.csv

• Generated synthetic claim_id values in the format CL101, CL102, etc.
• Injected a controlled fake fraudulent record to test risk scoring logic
• Created fraud detection logic based on:
  • High claim amount thresholds
  • Claim frequency
• Flagged high-risk and potentially fraudulent claims
• Final dataset saved as: Final_Insurance_Claims_Data.csv

• Created a stakeholder-facing dashboard using Tableau Desktop & Tableau Public
• Included KPIs and visual insights:
  • Total Claims
  • Total Fraud Claims
  • Percent Fraud
  • Top 5 Claim Submitters
  • Claims Distribution by Policy Type
  • Risk Breakdown by Flag Type

Recommendation for stakeholders:

• Investigate claimants with unusually high claim frequencies
• Review customers with high claim-to-premium ratios
• Use this dashboard as an early detection tool for fraud patterns

Insurance_Analytics/
├── data/
│   ├── raw/
│   │   ├── Insurance_Claims_Data.csv
│   │   └── Insurance_Customers_Data.csv
│   ├── processed/
│   │   ├── Cleaned_Merged_Insurance_Data.csv
│   │   ├── Feature_Engineered_Insurance_Data_with_ClaimID.csv
│   │   └── Final_Insurance_Claims_Data.csv
├── scripts/
│   ├── insurance_data_cleaning.py
│   ├── insurance_feature_engineering.py
│   ├── inject_fake_fraud_case.py
│   └── risk_scoring_fraud_detection.py
├── Insurance_Claims_Risk_Summary.png
└── README.md

This analysis examined 505 insurance claims to assess financial risk and detect potentially fraudulent activity using basic actuarial data science principles. Key findings:

• 5 claims (0.99%) were flagged as potentially fraudulent, primarily due to unusually high claim frequency or payout amounts.
• A synthetic fraud scenario was introduced to test our detection logic and validate feature engineering and risk scoring methods.
• A Tableau dashboard was created to visualize total claims, frequency patterns, and flagged risks designed to support underwriting, fraud review, or pricing teams.

• Monitor claimants with high claim-to-premium ratios or abnormal frequency.
• Investigate flagged claims and refine fraud thresholds based on future data volume.
• Integrate this process into monthly claims reporting to proactively manage financial exposure.

This project demonstrates how technical data workflows can be transformed into actionable insights for business and risk stakeholders.

• Tableau Public Dashboard