Multi-perspective analysis of D4D fine resolution data Movers - PowerPoint PPT Presentation

Multi-perspective analysis of D4D fine resolution data Movers Gennady & Natalia Andrienko, Georg Fuchs Trajectories M  (T  S) Fraunhofer Institute IAIS Spatial events Sankt Augustin Spatio-temporal positions E  (T  S) Germany Space (locations) Time (time units) http://geoanalytics.net/and Presence dynamics Spatial situations S  (T  P(M  E)) T  (S  P(M  E)) 1 G.Andrienko, N.Andrienko, G.Fuchs http://geoanalytics.net/and D4D challenge, Boston MA, 1 st of May, 2013

M ULTIPLE CONCEPTUAL VIEWS Spatial time ON MOVEMENT DATA series Aggregation Aggregation Extraction Extraction Integration Events Trajectories Extraction Movers Trajectories Locations Movement data Local time series Spatial events Spatial event data Spatial time series Times Spatial distributions 2 G.Andrienko, N.Andrienko, G.Fuchs http://geoanalytics.net/and D4D challenge, Boston MA, 1 st of May, 2013

E VALUATING THE DATA PROPERTIES Days with missing data 1. Change of IDs every two weeks 2. 3 G.Andrienko, N.Andrienko, G.Fuchs http://geoanalytics.net/and D4D challenge, Boston MA, 1 st of May, 2013

A SSESSING DAILY AGGREGATES FOR ANTENNAS Overall pattern: some days with missing data  4 G.Andrienko, N.Andrienko, G.Fuchs http://geoanalytics.net/and D4D challenge, Boston MA, 1 st of May, 2013

week A SSESSING DAILY AGGREGATES FOR ANTENNAS day of week Space-time pattern  - Systematically missing data for selected regions and time periods 5 G.Andrienko, N.Andrienko, G.Fuchs http://geoanalytics.net/and D4D challenge, Boston MA, 1 st of May, 2013

A NALYZING HOURLY AGGREGATES FOR ANTENNAS Yamoussoukro and San Pedro; raw counts (left) Vs. normalized (right)  hour day of week 6 G.Andrienko, N.Andrienko, G.Fuchs http://geoanalytics.net/and D4D challenge, Boston MA, 1 st of May, 2013

C LUSTERING ANTENNAS BY SIMILAR 1: High in evening: residential HOURLY TEMPORAL PROFILES districts, regular employment 2: Uniform calling activity: mix of residential and business 4: business districts 3,6,7: residential with partly employed population 7 G.Andrienko, N.Andrienko, G.Fuchs http://geoanalytics.net/and D4D challenge, Boston MA, 1 st of May, 2013

P EAK EVENT DETECTION AT ANTENNA LEVEL BASED ON HOURLY TIME SERIES 8 G.Andrienko, N.Andrienko, G.Fuchs http://geoanalytics.net/and D4D challenge, Boston MA, 1 st of May, 2013

A NALYSIS OF PEAKS : SIMULTANEOUS EVENTS IN 4 CITIES , UNUSUAL PROFILES 9 G.Andrienko, N.Andrienko, G.Fuchs http://geoanalytics.net/and D4D challenge, Boston MA, 1 st of May, 2013

F LOWS BETWEEN REGIONS THAT CORRESPOND TO PEAKS IN PEOPLE PRESENCE 10 G.Andrienko, N.Andrienko, G.Fuchs http://geoanalytics.net/and D4D challenge, Boston MA, 1 st of May, 2013

A NALYSIS OF FLOWS : DENSITY - DRIVEN V ORONOI TESSELLATION WITH R =100 KM 11 G.Andrienko, N.Andrienko, G.Fuchs http://geoanalytics.net/and D4D challenge, Boston MA, 1 st of May, 2013

S EMANTIC ANALYSIS OF PERSONAL PLACES One individual: home, work, social activities  12 G.Andrienko, N.Andrienko, G.Fuchs http://geoanalytics.net/and D4D challenge, Boston MA, 1 st of May, 2013

S EMANTIC ANALYSIS OF PERSONAL PLACES Work places and trajectories of  several persons in Abidjan 13 G.Andrienko, N.Andrienko, G.Fuchs http://geoanalytics.net/and D4D challenge, Boston MA, 1 st of May, 2013

C ONCLUSIONS We considered the data from multiple perspectives, including  - locations of varying resolution - time intervals of different length and hierarchical organization - trajectories We detected a number of interesting patterns that could facilitate a variety of  applications: - Reconstructing demographic information (to replace expensive and difficult to organize census studies) - Reconstructing patterns of mobility (to enhance transportation studies) - Identifying places of important activities (for improving land use and infrastructure) - Identifying events (for improving safety and security) - Detecting social networks (for marketing purposes) Data quality disables more sophisticated analyses  14 G.Andrienko, N.Andrienko, G.Fuchs http://geoanalytics.net/and D4D challenge, Boston MA, 1 st of May, 2013

M ONOGRAPH Springer, June 2013 ISBN 978-3-642-37582-8 397 p. 200 illus., 178 in colour Publisher’s leaflets with discount codes are available  15 G.Andrienko, N.Andrienko, G.Fuchs http://geoanalytics.net/and D4D challenge, Boston MA, 1 st of May, 2013