Exploring Taxi Cab Networks Over Time with a Dynamic Stochastic Block Model

• Activity Number: 174 - Dynamic Network Modeling
• Type: Contributed
• Date/Time: Monday, July 31, 2017 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistical Learning and Data Science
• Author(s): Beau Dabbs* and Giuliana Pallotta and Goran Konjevod
• Companies: Lawrence Livermore National Laboratory and Lawrence Livermore National Laboratory and Lawrence Livermore National Laboratory
• Keywords: Networks ; Dynamic ; Stochastic Block Model

Abstract

In this talk we examine the performance of a dynamic stochastic block model for weighted, directed networks. This work is motivated in part by the problem of modeling the network of taxi cab trips in NYC. For this dataset, various locations in the city are taken to be the nodes, and the number of trips from one node to another within a fixed time period are the weighted, directed edges in the network. The nature of this dataset requires models that are able to vary over time, since there are many day/night and weekly cycles within the dataset.

Our model also contains sender and receiver effects, allowing us to control for the inherently large degree of locations within Manhattan compared to locations in the other neighborhoods of the city. We first examine the ability of this model to estimate the patterns of movements within the city. Then we investigate the ability of this model to detect discrepancies caused by changes in the underlying network structure over time.