| Title |
Building a Traffic Accident Frequency Prediction Model at Unsignalized Intersections in Urban Areas by Using Adaptive Neuro-Fuzzy Inference System |
| Authors |
김경환(Kim, Kyung Whan) ; 강정현(Kang, Jung Hyun) ; 강종호(Kang, Jong Ho) |
| DOI |
https://doi.org/10.12652/Ksce.2012.32.2D.137 |
| Keywords |
비신호교차로;퍼지추론 시스템;뉴로-퍼지 ANFIS;unsignalized intersection;fuzzy inference system;neuro-fuzzy |
| Abstract |
According to the National Police Agency, the total number of traffic accidents which occurred in 2010 was 226,878. Intersection accidents accounts for 44.8%, the largest portion of the entire traffic accidents. An research on the signalized intersection is constantly made, while an research on the unsignalized intersection is yet insufficient. This study selected traffic volume, road width, and sight distance as the input variables which affect unsignalized intersection accidents, and number of accidents as the output variable to build a model using ANFIS(Adaptive Neuro-Fuzzy Inference System). The forecast performance of this model is evaluated by comparing the actual measurement value with the forecasted value. The compatibility is evaluated by R2, the coefficient of determination, along with Mean Absolute Error (MAE) and Mean Square Error (MSE), the indicators which represent the degree of error and distribution. The result shows that the $R^2$ is 0.9817, while MAE and MSE are 0.4773 and 0.3037 respectively, which means that the explanatory power of the model is quite decent. This study is expected to provide the basic data for establishment of safety measure for unsignalized intersection and the improvement of traffic accidents. |