Marco Veloso

Taxi-aware map: identifying and predicting vacant taxis in the city

Knowing where vacant taxis are and will be at a given time and location helps the users in daily planning and scheduling, as well as the taxi service providers in dispatching. In this paper, we present a predictive model for the number of vacant taxis in a given area based on time of the day, day of the week, and weather condition. The history is used to build the prior probability distributions for our inference engine, which is based on the naive Bayesian classifier with developed error-based learning algorithm and method for detecting adequacy of historical data using mutual information. Based on 150 taxis in Lisbon, Portugal, we are able to predict for each hour with the overall error rate of 0.8 taxis per 1×1 km2 area.

Urban mobility study using taxi traces

In this work, we analyze taxi-GPS traces collected in Lisbon, Portugal. We perform an exploratory analysis to visualize the spatiotemporal variation of taxi services; explore the relationships between pick-up and drop-off locations; and analyze the behavior in downtime (between the previous drop-off and the following pick-up). We also carry out the analysis of predictability of taxi trips for the next pick-up area type given history of taxi flow in time and space.

Sensing urban mobility with taxi flow

The analysis of taxi flow can help better understand the urban mobility. In this work, we analyze 177, 169 taxi trips collected in Lisbon, Portugal, to explore the relationships between pick-up and drop-off locations; the behavior between the previous drop-off to the following pick-up; and the impact of area type in taxi services. We also carry out the analysis of predictability of taxi trips given history of taxi flow in time and space.

Mining temporal patterns of transport behaviour for predicting future transport usage

information systems which are centred on the individual transport user. Especially, in dense urban cities where it is hard to oversee complex transport networks that are subject to frequent changes, maintenance and construction works, travellers want to be proactively notified about disruptions and traffic incidents relevant to their future behaviour. In this paper, we show how to mine characteristic patterns of the transport routines of urban bus riders for the design of novel travel information system that have the ability to understand forthcoming travel needs of individual users. We leverage on travel histories collected from automated fare collection system (AFC) to extract features of personal transport usage and study their predictive power to forecast whether people access public transport services on a future day or not.

Gaussian process-based predictive modeling for bus ridership

The dynamics of a city are characterized, among others, by the traveling patterns of its dwellers. Accurate knowledge of human mobility patterns would have applications, e.g., in urban design, in the optimization of public transportation operating costs, and in the improvement of public transportation services. The present paper combines a large scale bus transportation dataset with publicly available data sources to predict bus usage. We propose a Gaussian process-based approach for modeling and predicting bus ridership. To validate our approach we perform experiments on data collected from Lisbon, Portugal. The results demonstrate significant improvements in prediction accuracy compared to a probabilistic baseline predictor.

Catch Me If You Can: Predicting Mobility Patterns of Public Transport Users

Direct and easy access to public transport information is an important factor for improving the satisfaction and experience of transport users. In the future, public transport information systems could be turned into personalized recommender systems which can help riders save time, make more effective decisions and avoid frustrating situations. In this paper, we present a predictive study of the mobility patterns of public transport users to lay the foundation for transport information systems with proactive capabilities. By making use of travel card data from a large population of bus riders, we describe algorithms that can anticipate bus stops accessed by individual riders to generate knowledge about future transport access patterns. To this end, we investigate and compare different prediction algorithms that can incorporate various influential factors on mobility in public transport networks, e.g., travel distance or travel hot spots. In our evaluation, we demonstrate that by combining personal and population-wide mobility patterns we can improve prediction accuracy, even with little knowledge of past behavior of transport users.

Predictability of Public Transport Usage: A Study of Bus Rides in Lisbon, Portugal

This paper presents a study of the predictability of bus usage based on massive bus ride data collected from Lisbon, Portugal. An understanding of public bus usage behavior is important for future development of personalized transport information systems that are equipped with proactive capabilities such as predictive travel recommender systems. In this study, we show that there exists a regularity in the bus usage and that daily bus rides can be predicted with a high degree of accuracy. In addition, we show that there are spatial and temporal factors that influence bus usage predictability. These influential factors include bus usage frequency, number of different bus lines and stops used, and time of rides.

Exploring the relationship between mobile phone call intensity and taxi volume in urban area

As urbanization increases rapidly, there is a need for better understanding of the city and how it functions. Increasing digital data produced by the citys inhabitants holds great potential for doing so. In this work, an analysis of mobile phone call intensity and taxi volume in Lisbon, Portugal was carried out. With one source of data describes how city operates socially over mobile phone network and the other characterizes urban dynamic in traffic network, we discovered the inter-predictability between them. Based on one month of observation, we found that the variation in the amount of mobile phone calls was strongly correlated with the taxi volume of the previous two hours. Hence taxi volume can be used to predict mobile phone call intensity of the next two hours. In addition, we found that the level of inter-predictability varied across different time of the day; taxi was a predictor during PM hours while mobile phone call intensity became a predictor for taxi volume in AM hours. Strong correlations between these two urban signals were observed during active hours of the day and active days of the week.

A study on the suitability of GSM signatures for indoor location

Location is an important topic on Ambient Intelligence. Different techniques are used, alone or together, to determine the position of people and objects. One aspect of this problem concerns to indoor location. Various authors propose the analysis of Radio Frequency (RF) signatures as a solution for this challenge. An approach for indoor location is the use of RF signals acquired from a Global System for Mobile Communications (GSM) by Mobile Units(MU). In this paper we make a study based on around 485.000 signatures gathered from four buildings. We present our conclusions on the suitability and limitations of this approach for indoor location.

Visualization tool for taxi usage analysis: a case study of lisbon, portugal

This paper presents a visualization tool for taxi usage analysis. Data of taxi usage from Lisbon, Portugal is used as a case study. The tool operates on two modes; Mobility and Flow. Mobility mode displays an animation of taxi movement in the city with pick-up drop-off locations, as well as statistical information about the number of total trips made, current and recent available/occupied number of taxis, and top performance taxis. Mobility mode also allows the user to select any particular taxi to be displayed individually through a drop-down menu and search bar. Flow mode gives an overview of taxi movement with an animation of origin-destination (pick-up/drop-off) hourly flows, along with statistical graph of hourly trips made. The user can choose to view any particular time duration to observe the flow. The developed tool can be useful for taxi service providers in scheduling and dispatching management, as well as urban planning and design.