A Functional Data Analysis framework for spatial-temporal change detection

Authors: Avipsa Roy*, Arizona State University, Trisalyn A Nelson, University of California, Santa Barbara, Pavan Turaga, Arizona State University
Topics: Geographic Information Science and Systems, Temporal GIS, Transportation Geography
Keywords: Change detection; temporal alignment; functional data analysis, Strava, bicycling
Session Type: Virtual Paper
Day: 4/8/2021
Start / End Time: 8:00 AM / 9:15 AM
Room: Virtual 33
Presentation File: No File Uploaded

Monitoring change is an important aspect of understanding variations in spatial-temporal processes. Recently, big data on mobility, which are detailed across space and time, have become increasingly available. For example, bicycling data are available from crowdsourced apps such as Strava. New methods are needed to best utilize high spatial and temporal resolution for monitoring purposes, as data can be considered mappable time series, with varying sampling rates and issues of temporal misalignment. Our goal is to present a generalized framework for change detection from crowdsourced data after correcting for temporal misalignment using a functional data analysis framework. Our method enables analyzing big crowdsourced data captured continuously in time while addressing non-elastic rate variations in the underlying spatial-temporal processes. Using high resolution data from the Strava fitness app recorded every minute, we detect ridership changes in the Phoenix Metropolitan between 2017 and 2018. We quantified the changes across hourly and monthly scales in bicycling ridership volumes by street-segment to generate maps of change in Phoenix, AZ. Using spatially and temporally continuous data our study also advances the existing approaches to mobility analysis, by including all data about the underlying processes throughout the analysis, rather than monitoring change between discrete snapshots of time. Our method is reproducible by practitioners for monitoring changes in cities across multiple scales from crowdsourced ridership data and for making necessary infrastructure changes to assure the safety of bicyclists.

Abstract Information

This abstract is already part of a session. View the session here.

To access contact information login