• Rohan Yashraj Gupta
• Satya Sai Mudigonda
• Pallav Kumar Baruah

Fraud poses a major risk in any organizations due to its financial implications. To mitigate this, most companies have fraud detection models in place. This work, aim to study the use of Combined Actuarial Neural Networks in building a fraud detection model. In this, the classical generalized linear model (GLM) is embedded in a neural network model. This approach captures the data patterns which was not captured by using a classical generalized model. The dataset used for this study is publically available dataset “carclaims.txt’. It was observed that the combination of a classical generalized linear model with neural network models helped in increasing the sensitivity of the model. The highest sensitivity of 0.88 was observed when GLM was combined with a neural network containing embedding layers. Also, it was observed that the performance of all the models significantly improved once the oversampling of the dataset was done.

Automobile insurance, Data imbalance, Fraud detection, Machine learning, Neural networks, Actuarial models, Classification models

"Integrating actuarial models with neural networks for building a fraud detection model for automobile insurancex", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.6, Issue 6, page no.1138-1144, June 2019, Available :http://www.jetir.org/papers/JETIR1908D44.pdf

"Integrating actuarial models with neural networks for building a fraud detection model for automobile insurancex", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.6, Issue 6, page no. pp1138-1144, June 2019, Available at : http://www.jetir.org/papers/JETIR1908D44.pdf