This repository contains the full analytical pipeline for detecting potential fraud and anomaly patterns in obstetric (persalinan) claims within the Indonesian National Health Insurance (JKN) system.
The project is designed for:
- Internal BPJS Kesehatan research
- Health services evaluation
- Academic publication
To identify abnormal provider behavior patterns in obstetric claims using:
- Caesarean section rate anomalies
- Cohort-based utilization analysis
- Sequence of care reconstruction
- Supervised and unsupervised fraud modeling
Data source: BPJS Kesehatan claims database (BigQuery)
Pipeline:
Raw → Staging → Mart → Analytic Dataset → R Modeling → Publication Output
All transformation logic is version controlled.
01_sql
Contains all BigQuery SQL scripts used to construct staging, mart, and analytic datasets.
03_R
Contains data cleaning, feature engineering, modeling, and evaluation scripts.
04_outputs
Stores generated figures, tables, and model objects.
05_publication
Contains manuscript and supplementary material (RMarkdown).
This project uses:
- R
- renv (for dependency management)
- BigQuery SQL
To restore environment: