Real-time and Predictive Analytics for Smart Public Transportation Decision Support System

[pdf-embedder url=”https://wellapets.com/wp-content/uploads/2019/06/Real-time-and-Predictive-Analytics-for-Smart-Public.pdf” title=”Real-time and Predictive Analytics for Smart Public”]

Abstract—Public bus transit plays an important role in city transportation infrastructure. However, public bus transit is often difficult to use because of lack of real-time information about bus locations and delay time, which in the presence of operational delays and service alerts makes it difficult for riders to predict when buses will arrive and plan trips. Precisely tracking vehicle and informing riders of estimated times of arrival is challenging due to a number of factors, such as traffic congestion, operational delays, varying times taken to load passengers at each stop. In this paper, we introduce a public transportation decision support system for both short-term as well as long-term prediction of arrival bus times. The system uses streaming real-time bus position data, which is updated once every minute, and historical arrival and departure data – available for select stops to predict bus arrival times. Our approach combines clustering analysis and Kalman filters with a shared route segment model in order to produce more accurate arrival time predictions. Experiments show that compared to the basic arrival time prediction model that is currently being used by the city, our system reduces arrival time prediction errors by 25% on average when predicting the arrival delay an hour ahead and 47% when predicting within a 15 minute future time window.

Emerging trends and challenges. Bus systems are the backbones of public transit services in many cities. With their high capacity and relatively low investment and operational costs, bus systems can reduce traffic congestion substantially as well as bring environmental benefits such as reducing energy consumption and air pollution [1]. However, one major issue preventing many people from choosing bus service for commuting and travelling is its unpredictability [2]. Buses can often show up late due to various reasons: traffic congestion, road construction, special events or bad weather. This uncertainty forces potential riders to opt for other modes of
transportation.
Travel/arrival time prediction is one key research topic in intelligent transportation research [3], [4], [5]. Often, transit authorities use Automatic vehicle location (AVL) systems to monitor bus service status in order to provide information to city decision makers as well as commuters. The data collected provides the potential for more intelligent applications such as transit operation monitoring, smart trip planning, rough delay time estimation, at-stop displays, etc. In this paper, we focus on delay prediction for midsize cities. Midsize cities with populations of 10,000s-100,000s [6] of citizens. According to the statistics [7], more than 280 cities in the United States fall into the midsize city category.
Various statistical models have been applied to travel/arrival time prediction [3], [4], [5]. However, in most cases the system analyzed belonged to a large city with a large number of vehicle trips, generating a large dataset that was then used for analysis. Unlike large cities, midsize cities have limited public resources to invest in the transportation services, and moderate residential and employment density. As a result, the transit network is often not very dense and the vehicles are not scheduled very frequently. This reduces the amount of data that is available for creating prediction models, which can produce poor accuracy. In this paper, we shot that by using data available at the route segment level, it is possible to produce enough samples for more rigorous statistical analysis.
Contributions. This paper presents Transit-Hub, a decision support system that addresses the question of whether it is feasible to build a smart public transportation decision support system that can efficiently use utilize data from shared route segments to produce more accurate predictions. This paper’s main contributions are as follows:
∙ We present a clustering model that learns bus performance patterns during different hours of the day and different days of the week.
∙ We describe a real-time vehicle schedule adherence and prediction model. This model can be also used for identifying arrival time outliers and anomalous operations.
∙ We empirically validate our approach using a real-world dataset and real-time transit feed from Nashville. The experiments show that data collected over a two-hour window is most suitable for real-time prediction. Our
model provides a 25% reduction error in average arrival time prediction within the a 1-hour window and achieves a 47% improvement when predicting the delay for next 15 minutes compared to the model currently in use by the Nashville MTA.