For the given case study on Car Insurance Frauds, our team made a solution for detection of fraudulent transactions which used LightGBM along with centroid based under sampling.
The dataset can found in Dataset.xlsx file which has more than 10k+ genuine claims and around 640 fraudulent claims.
All the data visualization done for the project can found in the file Final EXL infographic.ipynb and certain important visualization done can be found in the Infographic.png
Details about the model and different models tried can be viewed in EXL-model.ipynb file.
The presentation for the finals can be found in the presentation file present in the repository.