Development of crash prediction model using artificial neural network (ANN): A case study of Hyderabad, India
By: Koramati, Siddardha
.
Contributor(s): Mukherjee, Arnab.
Publisher: USA Springer 2023Edition: Vol.104(1), Mar.Description: 63-80p.Subject(s): Humanities and Applied SciencesOnline resources: Click here
In:
Journal of the institution of engineers (India): Series ASummary: Machine learning (ML)-based prediction of urban crashes using secondary crash data derived from police crash databases is limited in India. This research demonstrated an artificial neural network (ANN) technique to formulate crash prediction models based on Hyderabad police data. A comprehensive crash database with data collected from Hyderabad police in 2015–2019 was used to develop a set of ANN models using various combinations of learning algorithms and activation functions. Among them, the ANN model with RMSProp and Sigmoid learning algorithm and Softmax and FTRL activation functions were observed to produce the best-fit prediction models. Based on a subsequent sensitivity analysis, “Cause of crash,” followed by “Road Geometry,” “Month of crash occurrence,” “Time of crash occurrence,” “crime vehicle type” and “victim vehicle type” were found to be sensitive factors influencing the crash outcome (fatal/non-fatal). Identification of such causative risk factors would help formulate targeted mitigation measures to improve the urban road safety status.
| Item type | Current location | Call number | Status | Date due | Barcode | Item holds |
|---|---|---|---|---|---|---|
Articles Abstract Database
|
School of Engineering & Technology Archieval Section | Not for loan | 2023-1758 |
Machine learning (ML)-based prediction of urban crashes using secondary crash data derived from police crash databases is limited in India. This research demonstrated an artificial neural network (ANN) technique to formulate crash prediction models based on Hyderabad police data. A comprehensive crash database with data collected from Hyderabad police in 2015–2019 was used to develop a set of ANN models using various combinations of learning algorithms and activation functions. Among them, the ANN model with RMSProp and Sigmoid learning algorithm and Softmax and FTRL activation functions were observed to produce the best-fit prediction models. Based on a subsequent sensitivity analysis, “Cause of crash,” followed by “Road Geometry,” “Month of crash occurrence,” “Time of crash occurrence,” “crime vehicle type” and “victim vehicle type” were found to be sensitive factors influencing the crash outcome (fatal/non-fatal). Identification of such causative risk factors would help formulate targeted mitigation measures to improve the urban road safety status.
Click on an image to view it in the image viewer
There are no comments for this item.