Predicting On-road Traffic Congestion from Public Transport GPS Data


  • Thura Kyaw
  • Nyein Nyein Oo
  • Win Zaw



GPS Data, Road Segments, Travel Speed Estimation Model, Traffic Jam Prediction Model, high prediction accuracy, Classifier Algorithms.


This paper predicts traffic congestion of urban road network by building machine learning models using public transport GPS
data. The bus GPS data are collected over 18 months started in September 2017 and ended in January 2019. After the data cleaning and data processing are carried out, time series data analysis is performed on these data. Travel Speed Estimation and
Traffic Jam Prediction Model are two major components of this work. Firstly, road network structure and GPS data sets are inputted to the Travel Speed Estimation Model to get estimated travel speed for every road segment in uniform time windows of a
day. The next step is to set up Traffic Congestion Prediction Model from estimated average travel speed and current GPS data
from the buses. Decision Trees, Random Forest Classifiers and ExtraTree Classifiers algorithms have been successfully applied
and validated with K-Fold cross validation yielding high prediction accuracy to a specific bus route in Yangon, Myanmar.




How to Cite

Thura Kyaw, Nyein Nyein Oo, & Win Zaw. (2020). Predicting On-road Traffic Congestion from Public Transport GPS Data . International Journal of Advances in Scientific Research and Engineering (IJASRE), ISSN:2454-8006, 6(3), 233–241.