Feature Engineering
• Features related to claim insurance with excellent performance in the classification of automatic payment, review, and investigation.

• Features that calculated the difference between the average billing amount by hospital and disease and each individual's billing amount play an important role.

Modeling :

• Three model ensembles with different random states set to complement the characteristics of extra trees with high randomness and increase stability.

• use of KNN, a case-based model

• a two-stage model was devised to classify automatic payment first, then review and investigation.

Evaluate the adequacy of insurance claims and propose recommendation services through customer type

• customer segmentation strategies and utilization plans were presented in the model
• cluster-based collaborative filtering, for disease prediction model based on the similarity of the customer's disease in the cluster.

Customer satisfaction analysis of All insurance company