Relationship research in spatial-time series of big data for forecasting road situation
Самарский национальный исследовательский университет им. акад. С.П. Королева, Ин-т информатики, математики и электроники, ф-т информатики, Каф. информационных систем и технологий, Россия, 443086, г. Самара, Московское шоссе, 34, Тел.: (846) 267-46-72, E-mail: firstname.lastname@example.org
Today, the technology stack is quite developed, which allows analyzing and displaying a situation that will unfold at some point in time on the basis of a large amount of heterogeneous data. However, in the field of road transport, existing tools operate with congestion and emergency situations on the road network, not covering other significant factors, such as: weather, cultural events, public behavior, etc. In addition, the forecast functions of existing tools aimed at a global short-term forecast of the situation, not allowing to localize possible problems.
The purpose of the work is to predict the development of the road traffic situation both on the entire street-road network and on its local section, based on the identified relationships in the spatio-temporal series of big data.
Considering that big data is constantly being updated, the ability to obtain results online to continuously increase the forecasting efficiency plays an important role, for which a software solution is proposed that allows, based on a client-server architecture, to perform the following research tasks:
- approximation of data on weather data sets, characteristics of traffic flows, traffic situation, congestion;
- clustering of sections of the road network into classes: fast, slow, high loaded, low loaded;
- identification of factors that reduce or increase the speed, density and intensity of traffic flows;
- modeling by groups of indicators and the conclusion of the corresponding results;
- forecasting the development of the road traffic situation.
Thus, the solution being developed will allow us to build a forecast of the road traffic situation for various indicators in a particular selected location based on many existing road, weather and organizational factors.