Font Size: a A A

Prediction Of The Severity Of Road Traffic Accidents Based On Spark Platform

Posted on:2021-03-22Degree:MasterType:Thesis
Country:ChinaCandidate:X G GuoFull Text:PDF
GTID:2492306230478264Subject:Software engineering
Abstract/Summary:PDF Full Text Request
Road traffic safety is closely related to people’s lives and has always been a major concern.Traffic accidents occur frequently every year and are not uncommon.Every traffic accident will bring people different degrees of impact and harm.Life and health and property safety pose serious hazards.Therefore,it is a great significance to study the severity of road traffic accidents,and propose measures to prevent or reduce and ameliorate the occurrence of traffic accidents,so as to reduce the degree of harm caused by traffic accidents to people accordingly.The volume of traffic accident data is relatively large.Traditional single-machine environments are limited by memory,CPU,and disk factors.Large or massive data cannot meet their storage requirements and processing speed requirements.In order to solve the problems and defects in processing large-scale data,this paper uses the Spark platform to have the advantages of processing speed and efficiency,and puts forward a method of predicting the severity of road traffic accidents based on Spark distributed platform,the method can effectively satisfy a larger quantity or mass data storage and processing requirements,improve the computational efficiency of the algorithm.This paper uses road traffic accident data,selects the characteristic variables that have a greater impact on the severity of traffic accidents,and constructs four predictive models of random forest,logistic regression,LightGBM,and MLP neural network on the Spark cluster platform.The four models are experimentally analyzed.The comparative analysis shows that LightGBM prediction model have a better effect on the prediction of the severity of traffic accidents,and has a higher accuracy.The LightGBM model has a slightly better prediction effect than the random forest model.Combine the LightGBM model and the random forest model to rank the importance of the characteristics that affect the severity of traffic accidents,and obtain the more important factors.These influencing factors are merged into human factors,vehicle factors,environmental factors,and road factors.Measures to prevent and ameliorate the occurrence of traffic accidents,and correspondingly reduce the harm caused by traffic accidents to people,these measures can provide reference and suggestions for roadinfrastructure and traffic management departments.
Keywords/Search Tags:Traffic accident severity prediction, Spark, Random forest, LightGBM, Neural Network
PDF Full Text Request
Related items