2021 OCT 04 (NewsRx) – By a Journalist-Staff News Editor at Daily Insurance News – Investigators discuss new findings in risk management. According to information published on Shanghai, People Republic of China, by the editors of NewsRx, the research said, “Safety performance functions (SPFs) are indispensable analytical tools that typically play a crucial role in estimating crash frequencies, identifying hot spots, analysis of factors contributing to accidents and evaluation of the effectiveness of safety countermeasures. Due to the limited availability of safety data, municipalities tended to adopt SPFs from the Highway Safety Manual or other neighboring jurisdictions.
Financial support for this research came from the International Program for Scientific and Technological Cooperation.
Our press reporters got a research citation from Ministry of EducationRecently, boosting algorithms have been exploited frequently for data analysis and statistical regression modeling problems. This research therefore aims to examine the effectiveness of improved calibration techniques to transfer SPF using limited regional data in an international context. To this end, AdaBoost.R2, an adaptive boost algorithm, TrAdaBoost.R2 in two stages, an instance-based transfer learning algorithm and a Gradient Boosting algorithm were used to study their efficiency in acquiring knowledge from source domain data available to predict crashes. in the target domain. For comparison, the calibration factor method was adopted to transfer the traditional negative binomial (NB) regression model. Two groups of training data sets were developed to train the four calibration techniques. The first group was used to examine the adaptability of the calibration techniques employed to the limited data of the target region. While the second group was used to further study the influence of broader vital information on the performance of the transferred models. This study was conducted between two we States, Florida and new York, and two Chinese cities, Shanghai and Suzhou. Based on the fit results, the amplification calibration techniques showed better prediction accuracy than the NB-based calibrated model using the limited target region data.
According to the editors, the research concluded: “Additionally, the amount and distribution of the training dataset were considered to be the two important factors that influence the proficiency of boosting calibration techniques.
This research has been peer reviewed.
For more information on this research, see: Applicability of hardening techniques in the calibration of safety performance functions for highways. Analysis and prevention of accidents, 2021; 159: 106193. Analysis and prevention of accidents can be contacted at: Pergamon-elsevier Sciences Ltd., The boulevard, Langford Way, Kidlington, Oxford OX5 1 GB, England. (Elsevier – www.elsevier.com; Analysis and prevention of accidents – http://www.journals.elsevier.com/accident-analysis-and-prevention/)
Our journalists point out that additional information can be obtained by contacting Xuesong wang, Ministry of Education, Key Lab Rd & Traff Engn, Shanghai, People Republic of China. Additional authors of this research include Ahmad Yehia, Mingjie Feng, Xiaohan Yang, Jian gong and Zhixing zhu.
The direct object identifier (DOI) for this additional information is: https://doi.org/10.1016/j.aap.2021.106193. This DOI is a link to an electronic document online that is free or to purchase, and can be your direct source for a journal article and its citation.
(Our reports provide factual information on research and discoveries from around the world.)