“…Methodologically, there are mainly two groups of approaches: statistical regression models and machine learning. The former includes the negative binomial regression model (Milton and Mannering, 1998;, the vector auto regression model (Wiebe et al, 2016), the cubic regression model (Moomen et al, 2018), the logistic regression model (Pugachev et al, 2017), multiple logistic regression (Hong et al, 2015), etc. In contrast to statistical methods based on prior assumptions of the input data, the principle of machine learning methods is to construct a nonlinear relationship between the input and output variables without any prior knowledge.…”