Relationship between Highway Geometric Characteristics and Accident Risk: A Multilayer Perceptron Model (MLP) Approach

Abstract

The traffic safety of mountain highway has always been one of the taking point. This study aims to collect road design data in large-scale research and analyzes the accident risk of highway geometric alignment. Accordingly, a method based on satellite maps and clustering algorithms is proposed to calculate the geometric alignment of the highway plane and its longitudinal section. The reliability of the method was verified on Nanfu highway in Chongqing, China. The planar and longitudinal sectional geometries of the four highways in Chongqing were obtained by the above method, and the corresponding 36,439 traffic accidents which occurred from 2010 to 2016 were used as the research objects. The accident risk of the highway geometry was analyzed based on the SHAP and MLP theories. The results show that the fitting and prediction abilities of the MLP model are better than those of the negative binomial model, and its correlation coefficient is improved by 33.2%. In addition, compared with the negative binomial model, the MLP model can estimate more accurately and flexibly the complex nonlinear relationship between the independent and the dependent variables.

A Study to Identify Long-Term Care Insurance Using Advanced Intelligent RST Hybrid Models with Two-Stage Performance Evaluation

“…(3) Machine learning techniques: Many studies have switched to machine learning techniques [22][23][24] to overcome the limitations of statistical assumptions [25] and successfully solve various prediction (or classification) problems in the real world. More common classification techniques used, such as rough set theory (RST) [26], DT [27], RF [28], and MLP [29,30], have become an important research trend at present. Moreover, Bayes network (BN) learning [31], logistic regression (LR) [32], naïve Bayesian (NB) [33], and support vector machine (SVM) [34] classifiers are always emerging techniques helpful for industry application fields; thus, they were also selected and emphasized in this study for the sake of comparison.…”

“…For example, the two results (with and without insurance) that are suitable for predicting LTCI are addressed in a method of reaction variable statistical model analysis on learning from the literature, and its past classification performance is also very outstanding. (3) The MLP neural network is one of the more popular neural networks at present [29,76] and has three layers, the input layer, hidden layer, and output layer, uses supervised learning, and can handle nonlinear problems with good classification performance. This neural network has the following advantages: high model accuracy, nonlinear model construction, different types of input variables, wide range of application fields, and ambiguity of input and output variables allowed.…”