🎯 Target Field (Label for Classification/Fraud Detection)

• Column: FraudFound_P
• Type: Binary (0 = Not Fraud, 1 = Fraud)
• This is your target field, used to identify whether a claim is fraudulent.

Dataset Link : https://www.kaggle.com/datasets/shivamb/vehicle-claim-fraud-detection

Python Pandas EDA Notebook : https://colab.research.google.com/drive/1ZYilF9HwXgAdiy8CRROsR8j559bzbjR2?usp=sharing

📊 Power BI Dashboard Ideas

1. Fraud Overview

• KPI Cards
  • Total Claims
  • Total Fraudulent Claims
  • Fraud Rate (%)
• Pie/Donut Chart
  • Fraudulent vs Non-Fraudulent Claims distribution

2. Demographics of Fraud

• Stacked Column/Bar Chart
  • Fraud by Sex, MaritalStatus, or AgeOfPolicyHolder
• Slicers for Age, Sex, Marital Status

3. Vehicle & Policy Analysis

• Heatmap or Clustered Bar Chart
  • Fraud by Make vs VehicleCategory
• Table
  • Top vehicle makes with highest fraud rate
• Bar Chart
  • Fraud by VehiclePrice, AgeOfVehicle, or PolicyType

4. Geographic Insight

• Map Visualization (if there were geographic data — but seems it’s missing)
• Otherwise:
  • Compare AccidentArea (Urban vs Rural) vs Fraud Rate

5. Agent & Claim Behavior

• Bar Chart
  • Fraud by AgentType (Internal vs External)
  • WitnessPresent vs Fraud Cases
  • PoliceReportFiled vs Fraud
• Trend Line or Area Chart
  • Claims per Month vs Fraud Cases (Month and MonthClaimed)

6. Policy Duration & Fraud

• Bar/Stacked Chart
  • Days_Policy_Accident, Days_Policy_Claim vs Fraud Rate