The goal of this project is to predict the potentially fraudulent providers based on the claims filed by them. Along with this, we will also discover important variables helpful in detecting the behaviour of potentially fraud providers. The primary goal was to leverage data wrangling, exploratory data analysis (EDA), preprocessing, and advanced modeling to enhance fraud detection capabilities within the healthcare sector.
- Data wrangling
- EDA
- Preprocessing
- Modeling
Health care fraud is a type of white-collar crime that involves the filing of dishonest health care claims to turn a profit. Healthcare fraud can affect anyone from healthcare funds to fund members like us. It happens when someone like a healthcare provider or an individual provides false information or withholds the information to get a financial benefit. Types of frauds by providers are: i) Billing for services that were not provided. ii) Duplicate submission of a claim for the same service. iii) Misrepresenting the service provided. iv) Charging for a more complex or expensive service than was provided. v) Billing for a covered service when the service provided was not covered.
Healthcare provider fraud poses significant challenges, affecting both the industry and patient outcomes. By harnessing the power of data science, we can enhance fraud detection mechanisms, protect resources, and ensure that healthcare services are delivered with integrity.
This case study aims to "predict the potentially fraudulent providers" based on the claims filed by them. Along with this, we will also discover important features helpful in detecting the behaviour of potentially fraudulent providers.
We can project the above problem to a binary classification problem to find whether the provider is fraud or non-fraud. Using the claim details submitted by that provider.
- 🔍 Data Wrangling: Unraveled complex healthcare datasets, ensuring optimal data quality and structure for analysis.
- 📊 EDA: Conducted in-depth exploratory data analysis to uncover patterns, anomalies, and potential indicators of fraud within the claims data.
- 🛠️ Preprocessing: Employed robust preprocessing techniques to clean, transform, and prepare the data for modeling, ensuring accuracy and reliability.
- 🤖 Modeling: Leveraged cutting-edge machine learning models to predict potentially fraudulent providers, contributing to the proactive identification of irregularities. Used Logistic Regression, Random Forest and Decision Tree models.