Deep Learning Ensemble Model for the Prediction of Traffic Accidents Using Social Media Data

Authors

  • Sudipta Sahana
  • Damodharan Palaniappan
  • Sunil Devidas Bobade
  • Shaik Mohammad Rafi
  • Kannadasan B
  • Jayapandian N

DOI:

https://doi.org/10.47750/pnr.2022.13.S09.055

Keywords:

traffic accidents, machine learning, ensemble learning, prediction model.

Abstract

We develop an enhanced accident occurrence prediction model which depends on the heterogeneous ensemble learning to tackle the topic of a accident period prediction in the early stages of the tragedy using millions of the traffic accident information’s from the India. In order to start with, we concentrate on the early stages of development of accidents and choose few useful data from five categories: location, the traffic, climate, objects, and the time field. Further, we implement data cleansing, processing of outlier, and the missing value of processing to raise the quality of the data. Data mining methods can support in foreseeing the factors that are influential in concern to make severe damages. The research has significant factors that are closely connected through the severity of accidents on thruways are identified by Random Forest. Top elements influencing unintentional seriousness include temperature, distance, wind Chills, moisture, direction of wind and visibility. The main aim of this research work is to give a architecture to anticipate road crashes gathering data from the social media handles and the open access data, by implementing a ensembled Deep Learning Model. After which the result shows decent outcomes as a resort to the problem and fulfills the objective of prediction model based on algorithms and deep Learning models.

Downloads

Published

2022-11-10

How to Cite

Sudipta Sahana, Damodharan Palaniappan, Sunil Devidas Bobade, Shaik Mohammad Rafi, Kannadasan B, & Jayapandian N. (2022). Deep Learning Ensemble Model for the Prediction of Traffic Accidents Using Social Media Data. Journal of Pharmaceutical Negative Results, 485–495. https://doi.org/10.47750/pnr.2022.13.S09.055

Issue

Section

Articles