Machine learning techniques for forecasting the traffic accident severity
| Autor: | Jose Arturo Garza-Reyes, Cherrafi Anass, Hmamed Hala, Benabbou Rajaa, Benghabrit Youssef |
|---|---|
| Rok vydání: | 2021 |
| Předmět: |
Receiver operating characteristic
Road traffic safety business.industry Computer science Reliability (computer networking) Decision tree Machine learning computer.software_genre Support vector machine Bayes' theorem Statistical classification Multilayer perceptron Artificial intelligence business computer |
| Zdroj: | 2021 International Conference on Digital Age & Technological Advances for Sustainable Development (ICDATA). |
| DOI: | 10.1109/icdata52997.2021.00018 |
| Popis: | The world cannot overstate the importance of road traffic safety, as it is an integral part of life. As consequence, there has been recently a marked advance in the use of machine learning techniques for the assessment of road traffic crashes. This study inspects the use of machine learning to build road safety model, by recognizing many factors that lead to accident severity, related to drivers, infrastructures, vehicles…. Different classification algorithms have been conducted to predict the severity of accidents based on real dataset, Decision Tree, Naives Bayes, Support Vector Machine, K-Nearest Neighbors, and Multilayer Perceptron. We compared the performance of each algorithm using the accuracy and the Receiver Operating Characteristic (ROC), to ensure that the proposed model provides stable and reliable predictive decisions. The finding revealed that the most accurate models are Support Vector Machine, K-Nearest Neighbors and Multilayer Perceptron with respectively 91% 92% 94% against the others models. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|