16. ABSTRACT Predictive models of urban mobility can help alleviate traffic congestion problems in future cities. State-of-the-art in travel demand forecasting is mainly concerned with long (months to years ahead) and very short term (seconds to minutes ahead) models. Long term forecasts aim at urban infrastructure planning, while short term predictions typically use high-resolution freeway detector/camera data to project traffic conditions in the near future. In this report, we present a medium term (hours to days ahead) travel demand forecast system. Our approach is designed to use cellular data that is collected passively, continuously and in real time to predict the intended travel plans of anonymized and aggregated individual travelers. The traffic conditions derived through traffic simulation can overcome the data sparsity for short term prediction. The data resolution, prediction tolerance and accuracy for medium term travel demand forecast are compromises between those of long term forecast and short term prediction.

Agent-based modeling in transportation problems requires detailed information on each of the agents that represent the population in the region of a study. To extend the agent-based transportation modeling with social influence, a connected synthetic population with both synthetic features and its social networks need to be simulated. However, either the traditional manually-collected household survey data (ACS) or the recent large-scale passively-collected Call Detail Records (CDR) alone lacks features. This work proposes an algorithmic procedure that makes use of both traditional survey data as well as digital records of networking and human behavior to generate connected synthetic populations. The generated populations coupled with recent advances in graph (social networks) algorithms can be used for testing transportation simulation scenarios with different social factors.

This paper explores the utility of peer pressure as an actionable mechanism to induce socially responsible and environmentally-conscious mobility habits. We adopt a two-stage game theoretic model of peer pressure to investigate feedback between social, geographic, and temporal dimensions of agent choices in a hyper-realistic micro-simulation of travel. The results show that peer pressure helps in achieving desirable equilibrium properties while reducing congestion and emissions due to sustained mode shift. With a way to initiate the required social norming and a proper concern for privacy and ethics, these cost-effective mechanisms may soon begin to find use in improving community welfare.

Locational data generated by mobile devices present an opportunity to substantially simplify methodologies and reduce analysis latencies in transportation planning applications. In this paper, we describe a modeling framework that supports most common transportation planning tasks, delivering actionable solutions at a fraction of time and cost as compared to the state of practice. Œe framework builds up on cell phone data processing and activitybased inferences of travel purposes with an Input-Output Hidden Markov Model (IO-HMM), followed by a Long Short Term Memory (LSTM) network that learns travelers’ mobility sequences. It combines the desired interpretability due to the parametric speci€cation of an IO-HMM with ƒexibility and predictive power of deep neural models. We describe our target use case for the synthesized activity chains: delivering decision support and transportation scenario evaluation to practitioners. We outline domain-driven operational objectives and verify that our framework meets these criteria by illustrating its usability in typical transportation demand planning applications. It is currently being deployed for testing by a major network carrier serving millions of users in the San Francisco Bay Area.

We proposed a novel FCN-ConvLSTM model to predict multi-focal human driver's attention merely from monocular dash camera videos. Our model has surpassed the state-of-the-art performance and demonstrated sophisticated behaviors such as watching out for a driver exiting from a parked car. In addition, we have demonstrated a surprising paradox: fine-tuning AlexNet on a largescale driving dataset degraded its ability to register pedestrians. This is due to the fact that the commonly practiced training paradigm has failed to reflect the different importance levels of the frames of the driving video datasets. As a solution, we propose to unequally sample the learning frames at appropriate probabilities and introduced a way of using human gaze to determine the sampling weights. We demonstrated the effectiveness of this proposal in human driver attention prediction, which we believe can also be generalized to other driving-related machine learning tasks.

Activity-based travel demand models are becoming essential tools used in transportation planning and regional development scenario evaluation. They describe travel itineraries of individual travelers, namely, what activities they are participating in, when they perform these activities, and how they choose to travel to the activity locales. However, data collection for activity based models is performed through travel surveys that are infrequent, expensive, and reflect the changes in transportation with significant delays. Thanks to the ubiquitous cell phone data, we see an opportunity to substantially complement these surveys with data extracted from network carrier mobile phone usage logs, such as call detail records (CDRs). In this paper, we develop input-output hidden Markov models to infer travelers' activity patterns from CDRs. We apply the model to the data collected by a major network carrier serving millions of users in the San Francisco Bay Area. Our approach delivers an end-to-end actionable solution to the practitioners in the form of a modular and interpretable activity-based travel demand model. It is experimentally validated with three independent data sources: aggregated statistics from travel surveys, a set of collected ground truth activities, and the results of a traffic micro-simulation informed with the travel plans synthesized from the developed generative model.

Urban modeling, including agent-based modeling of the coupled transportation and land use evolution, requires detailed information on each of the agents that represent the population in the region of a study. Traditional ways of obtaining this information include household surveys based on individual travel diaries. The surveys data provide a rich set of features, but they are limited in sampling size, geographical scope and frequency of updates. Moreover, they lack detail on inter-personal connections that give rise to social influences driving choice processes at a range of time scales. While manual surveying techniques are limited in their ability to collect social network data at scale, digital records of inter-personal communications provide an abundance of social networking information. This work proposes an algorithmic procedure that makes use of both traditional survey data as well as digital records of networking and human behaviours in generating connected synthetic populations for urban simulation.

Human decision making underlies data generating process in multiple application areas, and models explaining and predicting choices made by individuals are in high demand. Discrete choice models are widely studied in economics and computational social sciences. As digital social networking facilitates information flow and spread of influence between individuals, new advances in modeling are needed to incorporate social information into these models in addition to characteristic features affecting individual choices. In this paper, we propose two novel models with scalable training algorithms: local logistics graph regularization (LLGR) and latent class graph regularization (LCGR) models. We add social regularization to represent similarity between friends, and we introduce latent classes to account for possible preference discrepancies between different social groups. Training of the LLGR model is performed using alternating direction method of multipliers (ADMM), and training of the LCGR model is performed using a specialized Monte Carlo expectation maximization (MCEM) algorithm. Scalability to large graphs is achieved by parallelizing computation in both the expectation and the maximization steps. The LCGR model is the first latent class classification model that incorporates social relationships among individuals represented by a given graph. To evaluate our two models, we consider three classes of data to illustrate a typical large-scale use case in internet and social media applications. We experiment on synthetic datasets to empirically explain when the proposed model is better than vanilla classification models that do not exploit graph structure. We also experiment on real-world data, including both small scale and large scale real-world datasets, to demonstrate on which types of datasets our model can be expected to outperform state-of-the-art models.

Location based services and Geospatial web applications have become popular in recent years due to wide adoption of mobile devices. Search and recommendation of places or Points of Interests (PoIs) are prominent services available on them. The effectiveness of these services crucially depends on the availability of tags that are descriptive of places. The major geospatial databases that contain data about places suffer from the lack of descriptive tags for places, since writing them is a time-consuming process and only a few users do it despite having knowledge about places. In order to tackle this issue and automatically generate descriptive tags for places, we propose a solution that utilizes data about a set of events that happen in a specific place and use it to extract meaningful descriptive tags for that place. We use data about events held at places on Meetup, a well known event based social network and apply Latent Dirichlet Allocation (LDA) to derive sets of probable descriptive tags for any place. In order to evaluate our approach, we measure semantic relatedness between tags derived for places on Meetup and manually assigned tags from Foursquare, a location based service. Results show that event data can be used to derive semantically relevant place tags. This shows that location based services can benefit from capturing data about events to derive place tags.

ACM SIGSPATIAL workshop on Location Based Social Networks 2014 (http://faculty.ce.berkeley.edu/pozdnukhov/lbsn14/index.html) was held in conjunction with the 22nd ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems (SIGSPATIAL 2014) on November 4, 2014 in Dallas, Texas, USA. The objective of this workshop was to provide professionals, researchers, and technologists with a single forum where they can discuss and share the state-of-the-art of LBSN development and applications, present their ideas and contributions, and set future directions in emerging innovative research for location based social networks. This year program was composed of three sessions covering all aspects of LBSNs, with opening invited talks given by Dr. A. Haro (HERE/Nokia), Prof. M. Duckham (Uni Melbourne), Dr. Sen Xu (Twitter).

Large amount of time series data generated by sensors and Web users is great source of contextual information. Detecting outliers with unusually high values in time series data is crucial for inferring about any events in the real world. In this work, we describe an infinite Poisson mixture model to detect events by identifying outliers in time series of count data. This unsupervised technique estimates the probability densities of count data which have an unknown Poisson mixture while it simultaneously detects outliers in the data. The advantage of our model is that outliers are mapped to mixture components discovered by infinite mixture model and thus inference can be drawn on the different 'types' of outliers and their proportions in the data. This lets us identify and categorize events based on magnitude of outlier data. We have analysed the performance of our model against a well known event detection technique based on Markov Modulated Poisson Process (MMPP) using synthetic and real world data. Results show that our approach to detecting events is more appropriate in analysing periodic count data as compared to the MMPP baseline. The experiments demonstrate that the presented model provides robust, detailed, and interpretable results for the analysis of outliers to detect events.

A new convex optimization framework is developed for the route flow estimation problem from the fusion of vehicle count and cellular network data. The issue of highly underdetermined link flow based methods in transportation networks is investigated, then solved using the proposed concept of cellpaths for cellular network data. With this data-driven approach, our proposed approach is versatile: it is compatible with other data sources, and it is model agnostic and thus compatible with user equilibrium, system- optimum, Stackelberg concepts, and other models. Using a dimensionality reduction scheme, we design a projected gradient algorithm suitable for the proposed route flow estimation problem. The algorithm solves a block isotonic regression problem in the projection step in linear time. The accuracy, computational efficiency, and versatility of the proposed approach are validated on the I-210 corridor near Los Angeles, where we achieve 90% route flow accuracy with 1033 traffic sensors and 1000 cellular towers covering a large network of highways and arterials with more than 20,000 links. In contrast to long-term land use planning applications, we demonstrate the first system to our knowledge that can produce route-level flow estimates suitable for short time horizon prediction and control applications in traffic management. Our system is open source and available for validation and extension.

This article introduces a microsimulation of urban traffic flows within a large-scale scenario implemented for the Greater Dublin region in Ireland. Traditionally, the data available for traffic simulations come from a population census and dedicated road surveys that only partly cover shopping, leisure, or recreational trips. To account for the latter, the presented traffic modeling framework exploits the digital footprints of city inhabitants on services such as Twitter and Foursquare. We enriched the model with findings from our previous studies on geographical layout of communities in a country-wide mobile phone network to account for socially related journeys. These datasets were used to calibrate a variant of a radiation model of spatial choice, which we introduced in order to drive individuals’ decisions on trip destinations within an assigned daily activity plan. We observed that given the distribution of population, the workplace locations, a comprehensive set of urban facilities, and a list of typical activity sequences of city dwellers collected within a national travel survey, the developed microsimulation reproduces not only the journey statistics such as peak travel periods but also the traffic volumes at main road segments with surprising accuracy.

Mobility data has increasingly grown in volume over the past decade as localisation technologies for capturing mobility flows have become ubiquitous. Novel analytical approaches for understanding and structuring mobility data are now required to support the backend of a new generation of space-time GIS systems. It is increasingly important as GIS is becoming a decision support platform for operations in fleet management, urban data analysis and related applications. This paper applies the machine learning method of probabilistic topic modelling for semantic enrichment of mobility data recorded in terms of trip counts by using geo-referenced social media data. It further explores the questions of causality and correlation, as well as predictability of the obtained semantic decompositions of mobility flows on a real dataset from a bike sharing network.

Location-based social networks serve as a source of data for a wide range of applications, from recommendation of places to visit to modelling of city traffic, and urban planning. One of the basic problems in all these areas is the formulation of a predictive model for the location of a certain user at a certain time. In this paper, we propose a new approach for predicting user location, which uses two components to make the prediction, based on (i) coordinates and times of user check-ins and (ii) social interaction between different users. We improve the performance of a state-of-the art model using the radiation model of spatial choice and a social component based on the frequency of matching check-ins of user's friends. Friendship is defined by the presence of reciprocal following on Twitter. Our empirical results highlight an improvement over the state-of-the-art in terms of accuracy, and suggest practical solutions for spatio-temporal and socially-inspired prediction of user location.

We present a methodology to measure multi-modal interconnectivity between different transportation modes that operate in a city. The interconnectivity of an urban network represents how well different services integrate to offer seamless transportation options to users. On the supply side, we leverage open data sources that cities provide to accurately model the services that they offer. On the demand side, we account for myopic user behavior through the use of journey planners and travel demand estimates. The reciprocal interaction between supply and demand is then used to characterize interconnectivity. In a multi-modal setting, we present different measures that can be employed to understand shortcomings in connectivity across the network. Using these metrics, improvements in service schedules are proposed using an optimization model that seeks to perturb existing schedules to improve transit connectivity. Real-world data from Washington, D.C. is used to demonstrate presented measures and optimal schedule perturbation for one route are presented which result in a 19% reduction in delays.

In this paper, we develop a data-driven methodology to characterize the likelihood of orographic precipitation enhancement using sequences of weather radar images and a digital elevation model (DEM). Geographical locations with topographic characteristics favorable to enforce repeatable and persistent orographic precipitation such as stationary cells, upslope rainfall enhancement, and repeated convective initiation are detected by analyzing the spatial distribution of a set of precipitation cells extracted from radar imagery. Topographic features such as terrain convexity and gradients computed from the DEM at multiple spatial scales as well as velocity fields estimated from sequences of weather radar images are used as explanatory factors to describe the occurrence of localized precipitation enhancement. The latter is represented as a binary process by defining a threshold on the number of cell occurrences at particular locations. Both two-class and one-class support vector machine classifiers are tested to separate the presumed orographic cells from the nonorographic ones in the space of contributing topographic and flow features. Site-based validation is carried out to estimate realistic generalization skills of the obtained spatial prediction models. Due to the high class separability, the decision function of the classifiers can be interpreted as a likelihood or susceptibility of orographic precipitation enhancement. The developed approach can serve as a basis for refining radar-based quantitative precipitation estimates and short-term forecasts or for generating stochastic precipitation ensembles conditioned on the local topography.

This paper explores the community structure of a network of significant locations in cities as observed from location-based social network data. We present the findings of this analysis at multiple spatial scales. While there is previously observed distinct spatial structure at inter-city level, in a form of catchment areas and functional regions, the exploration of in-city scales provides novel insights. We present the evidence that particular areas in cities stratify into distinct “habitats” of frequently visited locations, featuring both spatially overlapping and disjoint regions. We then quantify this stratification with normalized mutual information which shows different stratification levels for different cities. Our findings have important implications for advancing models of human mobility, studying social exclusion and segregation processes in cities, and are also of interest for geomarketing analysts developing fidelity schemes and promotional programmes.

An algorithmic architecture for kernel-based modelling of data streams from city sensing infrastructures is introduced. It is both applicable for pre-installed, moving and extemporaneous sensors, including the “citizen-as-a-sensor” view on user-generated data. The approach is centred around a kernel dictionary implementing a general hypothesis space which is updated incrementally, accounting for memory and processing capacity limitations. It is general for both kernel-based classification and regression. An extension to area-to-point modelling is introduced to account for the data aggregated over a spatial region. A distributed implementation realised under the Map-Reduce framework is presented to train an ensemble of sequential kernel learners.