- Dataset : https://www.kaggle.com/datasets/carlosalvite/car-insurance-claims-event-log-for-process-mining?resource=download
- This event log has been artificially generated and curated to provide a comprehensive view of car insurance claims, allowing users to discover and identify bottlenecks, automation opportunities, conformance issues, reworks, and potential fraudulent cases using any process mining software.
Standard Process flow: First Notification of Loss (FNOL) → Assign Claim → Claim Decision → Set Reserve → Payment Sent → Close Claim
- Goal : a Fraud detection system, combining the power of process mining analysis and machine learning. Analyzed event logs to identify suspicious patterns, then predicted fraud likelihood using ML.
| Feature Name | Description | Type | Example |
|---|---|---|---|
| case_id | A unique identifier for each insurance claim case. | Categorical | 000112d5-9d04-450f-820f-3edfc0626cf9 |
| client | The name of the client involved in the claim. | Categorical | Kimberly Collins |
| agent | The name of the insurance agent assigned to the claim. | Categorical | Hannah Lopez |
| adjuster | The name of the adjuster responsible for handling the claim. | Categorical | Joy Simpson |
| claim_amount | The monetary value of the insurance claim. | Numerical | 9266.19 |
| client_age | The age of the client involved in the claim. | Numerical | 60 |
| type_of_policy | The type of insurance policy covering the claim. Possible values include: Comprehensive, Collision, etc. | Categorical | Comprehensive |
| car_brand | The brand of the car involved in the accident. | Categorical | Hyundai |
| car_model | The model of the car involved in the accident. | Categorical | Elantra |
| car_year | The manufacturing year of the car involved in the accident. | Numerical | 2021 |
| type_of_accident | The type of accident involved in the claim. Possible values include: Head-on, Rear-end, Side-impact, Rollover, etc. | Categorical | Head-on |
| user_type | Indicates whether the user handling the claim is a human or an RPA (Robotic Process Automation). | Categorical | Human |
| total_activities | The total number of activities recorded for the claim process. | Numerical | 6 |
| throughput_duration_days | The total duration of the claim processing in days. | Numerical | 31 |
| First Notification of Loss (FNOL) | The sequential order of the activity "First Notification of Loss" in the claim process. | Numerical | 1 |
| Assign Claim | The sequential order of the activity "Assign Claim" in the claim process. | Numerical | 2 |
| Claim Decision | The sequential order of the activity "Claim Decision" in the claim process. | Numerical | 3 |
| Set Reserve | The sequential order of the activity "Set Reserve" in the claim process. | Numerical | 4 |
| Payment Sent | The sequential order of the activity "Payment Sent" in the claim process. | Numerical | 5 |
| Close Claim | The sequential order of the activity "Close Claim" in the claim process. | Numerical | 6 |
| State | The final status of the claim, indicating whether it was valid, fraudulent or violation. Possible values include: valid, fraud, violation. | valid |
- Celonis
- python & jupyter notebook