|
Predictions of the Probability of Winter Traffic Accidents
|
|
| Author:
|
SHAIK.MAHABOOB BASHA, NIDAMANURI.SRINU, J.SASI KIRAN, K.M.RAYUDU, V.MADHAV REDDY
|
| Abstract:
|
The number of people killed in car accidents due to weather is higher in the winter. The majority of road accidents are caused by human mistake, defective equipment, inadequate infrastructure, or adverse weather. With a more precise method of estimating the likelihood of road accidents in reaction to changes in road weather, we may save lives. Data from traffic accident reports between 2017 and 2019 were used in the study. The data imbalance problem between weather-related and other types of road accidents was solved using the synthetic minority oversampling technique (SMOTE), road geometry and accident area altitude were determined using Data from geographic information systems (GIS) and the Shuttle Radar Topography Mission (SRTM) were fed into a machine learning pipeline where various models (random forest, XGBoost, neural networks, etc.) were used. The data was split 7:3 between a training set and a test set. In order to categorise the severity of traffic accidents in relation to weather hazards, the random forest model showed the best results. Therefore, it may be possible to provide a service that predicts the likelihood of road accidents owing to weather conditions in the winter by incorporating meteorological data and road geometry into machine learning algorithms.
|
| Keyword:
|
SMOTE, Spatial Interpolation, Machine Learning, and the Traffic Safety Service.
|
| EOI:
|
-
|
| DOI:
|
https://doi.org/10.31838/ijpr/2019.11.03.170
|
| Download:
|
Request For Article
|
|
|