We are happy to invite to two open colloquium talks by Dr. Eric Delmelle, Associate Professor, Department of Geography & Earth Sciences, University North Carolina Charlotte about the two interesting topics:
“Evaluating Travel Impedance Agreement among Online Road Network Data Providers” & “Visualizing the dynamics of health-related tweets: opportunities and computational challenges”
The talks will be on Wednesday Feb 17, 2016 4.15 pm, Hörsaal Berliner Straße 48, Institute of Geography, Heidelberg University.
Further information can be found here
Evaluating Travel Impedance Agreement among Online Road Network Data Providers
Online mapping providers offer unprecedented access to spatial data and analytical tools; however the number of analytical queries that can be requested is usually limited. As such, providers using Volunteered Geographic Information (VGI) offer a viable alternative, given that the quality of the underlying spatial data is adequate. In this presentation, I will present results of an analysis aimed at assessing the agreement in travel distance estimates between Mapquest Open–which embraces OpenStreetMap (OSM) data, a VGI dataset-, and two other popular commercial providers, namely Google Maps™ and ArcGIS™ Online. We use a routing service Application Program Interface (API), to estimate travel impedance and the average number of OSM contributors. Origin-destination pairs are simulated for the state of North Carolina, U.S.A., and travel estimates reported for each of the providers. Results suggest (1) a strong correlation among all three road network providers, (2) agreement improves with increasing route distances and (3) decreases in areas with denser road network as providers may select different routes from a larger number of potential paths. Most importantly, travel estimates from Mapquest Open exhibit stronger similarity with both commercial providers when the average number of OSM contributors along the selected path is larger.
Visualizing the dynamics of health-related tweets: opportunities and computational challenges
Twitter, a form of social media, provides endless opportunities for public health. Information on disease symptoms, generated by twitter users, may alert health officials on the risk posed by a certain disease before it can be detected and officially confirmed in a lab. Certain tweets are georeferenced, and coupled with their temporal stamps, they have the potential to be used for space-time monitoring of diseases. This is particularly important to better understand disease dynamics, such as seasonality, direction, intensity and risk of diffusion to new regions. In this presentation, I will focus on the computational challenges associated to infer meaningful information from twitter related data, with an application to pollen-related tweets. I will then discuss a space-time framework to visualize the intensity of collected tweets in both space and time. I will discuss the impact of positional, temporal and attribute accuracy on the detection of space-time clusters from health-related tweets.