Using visualization and predictive analysis to predict train delays

Sara Lbazri,Younes Oubrahim,Mohamed Azouazi,Amina Rachik

doi:10.21533/pen.v6i2.537

Abstract

France has the second largest European railway network, with a total of 29,901 kilometers of railway. However, the travel experience of passengers is frequently marked by delays, late arrival of trains at stations, causing inconvenience. The purpose of this paper is to present a new approach for visual prediction of train delays. Our approach is driven by predictive analysis and interactive visualization. The study has benefitted from access to open data SNCF including information about train delays , train number , station , departure and arrival time .Based on this data we develop a new workflow for predictive analysis including visualization in all steps from data pre-processing to deployment .

Highlights

Predictive analytics is an important part of data analysis, dealing with the problem of the quantitative or qualitative evaluation of the results of a given process [1]
Predictive analytics has been successfully applied in many applications areas as expert tools, they usually require a good choice of models, parameters and good quality of learning data
The SNCF have put in the hands of decision makers a public datasets containing information on trains, delays, schedules, information on stations, to motivate datascients to improve the field of French rail transport [2]

Summary

Introduction

Predictive analytics is an important part of data analysis, dealing with the problem of the quantitative or qualitative evaluation of the results of a given process [1]. Predictive analytics has been successfully applied in many applications areas as expert tools, they usually require a good choice of models, parameters and good quality of learning data. The evaluation of prediction uncertainties remains difficult, especially if prediction models are applied in the black box manner. Our idea is to use visualization in predictive analytics, in order to improve the forecasting process and enable the human to interact throughout the process. We will propose a new approach to predict train delays using visualization and predictive analysis. The contributions of this article is to propose a new workflow for predictive analysis including visualization techniques and the human in the predictions of train delays

Literature survey

Findings

CONCLUSION