MODAP Clinic: ProAnVis - Processing, Analysis & Visualisation of Tracking Data

As an active MODAP member, I must say that MODAP Clinic has indeed benefited me as a Phd student and also as a young research. Below is a summary of workdone by sub-group formed to explore gps-based data.

Group 2 research focus: Exploration of LISA with Pedestrian data set – Rotterdam case study

The group research focus was derived from an impressive presentation by Arend Ligtenberg, who is an Assistant Professor in Spatial Analysis and Modelling at Wageningen University, during the MODAP Clinic workshop. Participants interested in exploring Local Index of Spatial Association (LISA) concept with pedestrian data from Rotterdam formed this group and worked with Arend.  The team members were Godwin Yeboah from Northumbria University; Steffie Vannieuland from Ghent University; Paolo Cintia from University of Pisa; Joris van den Bossche from Ghent University; and  Lorenzo Gabrielli from University of Pisa.

Using the cluster and outlier analysis too, LISA tool in ArcGIS10 software, and gps data of pedestrians from Rotterdam, spatial clusters of point features – representing pedestrian movements – with recorded speed as attribute were identified. The output from the application of the tool comprises Local Moran’s I value, a p-value, a z-score, and a cluster type of each point feature.  We concentrated on cluster type labelled LL which indicated point features with high statistical significance of spatial association and low speed values – identified stops in the data. Also, HH which indicated High speed values – identified non-stops/passes in the data were visualised together using OpenStreetMap (OSM) as a base-map. Thereafter, LISA Z Score versus Movement speed scatterplot was generated using MS Excel and interpreted. Additionally, extracted stops were further explored vis-à-vis number of trips per stop as well as median number of points per person per stop.  We also investigated the quality of the speed in the raw data by plotting calculated speed versus GPS speed using scatterplots and Boxplot – resulting in non-convergence.

As a way of concluding, the tool was very easy to use in ArcGIS10 but quite slow with large datasets – limiting the group to use only BK Parking data – WestBlaak Parking. We found the largest stop cluster around the area between Cinerama and Keerweer street with the cluster cloud covering the whole segment of Westblaak street segment; the second was in and around Lijnbaan parking area. Calclulated speed against GPS speed did not converge. The group did a twenty minutes presentation of workdone to all participants and students/team from Urbanism/Department of Architecture who collected the Rotterdam data. Kudos to MODAP Clinic organizers – we have been well diagnosed and the prescriptions are worth a good attention!

Further details of the workshop can be found here.

Note:  - Please note that the words are my opinion and do not mean any harm to anyone.