Application of Deep Learning and Neural Network to Speeding Ticket and Insurance Claim Count Data

With the popularity of big data analysis with insurance claim count data, diverse regression models for response variable have been developed. However, there is a multicollinearlity issue multivariate input variables to models. Recently, deep learning and neural network proposed, Keras Tensorflow-based model has also proposed. To apply non-normal we perform root mean square error accuracy comparison gradient boosting machines (a popular machine tree algorithm), principal component (PCA)-based Poisson regression, PCA-based negative binomial zero inflated poisson avoid multicollinearity simulated normal distribution combined normally distributed binary copula-based asymmetrical two real sets, which consist speeding ticket Singapore data.