Authors: Krasen Samardzhiev*, University of Liverpool
Topics: Geographic Information Science and Systems, Temporal GIS, Quantitative Methods
Keywords: GIS, urban data, time series, clustering
Session Type: Paper
Start / End Time: 1:10 PM / 2:50 PM
Room: Congressional A, Omni, West
Presentation File: No File Uploaded
A significant part of human interactions happen through mobile phone calls, text messages or apps. The nature of these communication channels enables the capture of large amounts of data, that can later be analysed to gain various insights.
The goal of this paper is to find similar groups of areas within a city, where the similarity is based on mobile phone activity patterns across time.
To that end new methods from topological data analysis (TDA), a new and expanding field, focused on applying insights from mathematical topology to data analysis, will be used. TDA techniques focus on finding the underlying 'shape' of the space the data resides in, and have been successfully used in various domains such as medicine, chemistry and finance. Specifically, the focus is on developing a new clustering method that will take into account the 'topological' properties and patterns of the phone activity within an area.
In order to access the effectiveness of the method, phone activity data from the city of Milan will be used. Preliminary results suggest that the method captures and takes into account properties of an area's activity pattern, such as how cyclic or disconnected it is, which are ignored by methods such as k-means clustering.
 Telecom Italia, 2015, "Telecommunications - SMS, Call, Internet - MI", https://doi.org/10.7910/DVN/EGZHFV, Harvard Dataverse, V1