Vehicle Claim

Dataset Overview

Dataset List

1. Vehicle Claim - Synthetic dataset created using the DVI dataset.
2. Car Insurance - Dataset from Kaggle, link: Car Insurance.
3. Vehicle Insurance - Dataset from Github, link: Vehicle Insurance.

Vehicle Claim dataset details

• Creation code: Code to create the dataset.
• Dataset storage location: Dataset storage location.
• Attribute list:
  • Maker - Categorical, vehicle make.
  • GenModel - Categorical, vehicle model.
  • Color - Categorical, vehicle color.
  • Reg_Year - Categorical, registration year.
  • Body_Type - Categorical, e.g., SUV, Convertible.
  • Runned_Miles - Numerical, vehicle mileage.
  • Engin_Size - Categorical, engine size.
  • GearBox - Categorical, automatic or manual.
  • FuelType - Categorical, gasoline or diesel.
  • Price - Numerical, vehicle price.
  • Seat_num - Numerical, number of seats.
  • Door_num - Numerical, number of doors.
  • issue - Categorical, damage type.
  • issue_id - Categorical, specific damage.
  • repair_complexity - Categorical, repair difficulty.
  • repair_hours - Numerical, repair time required.
  • repair_cost - Numerical, repair cost.

Training and Evaluation Parameters

• Training parameters:
  • dataset - Training dataset selection (vehicle_claims, car_insurance, vehicle_insurance).
  • data - Data type (normal or mixed).
  • encoding - Categorical feature encoding method.
  • numerical - Whether to use only numerical features.
  • batch_size - Batch size.
  • epoch - Number of training epochs.
  • latent_dim - Latent space dimension.
• Evaluation parameters:
  • threshold - Evaluation threshold.

Citation

• Paper citation:

@article{ Author = {Ajay Chawda and Stefanie Grimm and Marius Kloft}, Title = {Unsupervised Anomaly detection for Auditing Data and Impact of Cetgorical Encodings}, Journal = {https://arxiv.org/abs/2210.14056}, Year = {2022}, }