DETECTING OUTLIER IN HEALTH INSURANCE CLAIM DATA USING MACHINE LEARNING
Health insurance is a form of protection that offers coverage for medical expenses to policyholders according to the agreements stated in the insurance policy. One of the challenges faced by insurance companies is the handling of unusual or outlier claims. Machine learning can be employed to predict...
Saved in:
| Main Author: | |
|---|---|
| Format: | Theses |
| Language: | Indonesia |
| Online Access: | https://digilib.itb.ac.id/gdl/view/80599 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Institut Teknologi Bandung |
| Language: | Indonesia |
| Summary: | Health insurance is a form of protection that offers coverage for medical expenses to policyholders according to the agreements stated in the insurance policy. One of the challenges faced by insurance companies is the handling of unusual or outlier claims. Machine learning can be employed to predict the occurrence of outliers. Three machine learning classification models are utilized to address this issue: Random Forest, Naïve Bayes, and Support Vector Machines. The desired prediction outcomes sometimes involve not only hard classification but also a soft classification with probabilities value. Here, our model is conducted in two ways: one that produces binary class outputs and another that produces probability outputs. Calibration is necessary for models with probability outputs to ensure accurate predictions. The calibration methods employed include Platt Scaling, Isotonic Regression, and Beta Calibration. The models are then applied to two data groups, namely inpatient and outpatient data. Simulation results indicate that the Random Forest model outperforms the other models. |
|---|