PULSE: A Real Time System for Crowd Flow Prediction at Metropolitan - - PowerPoint PPT Presentation
PULSE: A Real Time System for Crowd Flow Prediction at Metropolitan Subway Stations Ermal Toto, Prof. Elke A. Rundensteiner Prof. Yanhua Li 1 Outline Introduction Challenges State of the art Proposed Solution Experimental
1
2
3
United Nations. (2014). World Urbanization Prospects 2014: Highlights. United Nations Publications.
4
transportation networks
Annez, P. C., & Buckley, R. M. (2009). Urbanization and growth: setting the context. Urbanization and growth, 1, 1-45.
5
6
7
8
Subway Stations Bus Stations
9
10
Stathopoulos, A., & Karlaftis, M. G. (2003). A multivariate state space approach for urban traffic flow modeling and prediction. Transportation Research Part C: Emerging Technologies, 11(2), 121-135. 11
t t-1 t-2 t-3 t+1
multivariate by handcrafting transfer functions that model the interaction between different variables in a regression model.
many nodes. Prediction Current Time
Sun, H., Liu, H. X., Xiao, H., He, R. R., & Ran, B. (2003, January). Short term traffic forecasting using the local linear regression model. In 82nd Annual Meeting of the Transportation Research Board, Washington, DC.
Clark, S. (2003). Traffic prediction using multivariate nonparametric regression. Journal of transportation engineering, 129(2), 161-168. 12
Features: Local Streams, Remote Streams, Weather, Time Information, etc.
manually, therefore models are not scalable. Prediction Model Prediction at t+i
Hamner, B. (2010, December). Predicting travel times with context-dependent random forests by modeling local and aggregate traffic flow. In Data Mining Workshops (ICDMW), 2010 IEEE International Conference on (pp. 1357-1359). IEEE.
Vlahogianni, E. I., Karlaftis, M. G., & Golias, J. C. (2005). Optimized and meta-optimized neural networks for short-term traffic flow prediction: a genetic approach. Transportation Research Part C: Emerging Technologies, 13(3), 211-234. 13
Prediction Model Features: Local Streams, Remote Streams, Weather, Time Information, etc. Prediction at t+i
manually, therefore models are not scalable.
14
15
16
17
18
Average number of arrivals to a station per (15min) interval. Average time duration of trips arriving to a station (from any station).
Attrition Rate is the ratio of trips that are departing only (do not have a matching return trip). Peak scores capture peak traffic behaviors during mornings and evenings, for both arrivals and departures. They are defined by the number of local outliers.
Horizon = 4 Time Based Stream Selection is based on the assumption that future arrivals at a station, will come from departures of other stations that are within the prediction horizon.
Horizon =12 Time Based Stream Selection is based on the assumption that future arrivals at a station, will come from departures of other stations that are within the prediction horizon.
Flow Based Stream Selection is based on the assumption that future arrivals at a target station will come from stations with high historical traffic to that station.
Prediction Actual Flow Time Interval
27
28
29
─ Days and Nights ─ Holidays, Fridays etc ..
─ A locations behavior during weekends, is assumed to have no relation to its behavior during the weekdays, unless otherwise described by the personality features.
30
Brute Force:
Values x 118 FBSS Values x 118 PFBSS Values ~ 2.3Billion Models
443.7years (For each ML Method, therefore x 5).
14.7years : ).
126Million Years 14.7 Years
31
Personality Features, Horizon, Day of the Week Classification Model Using KNN KNN, RF, ANN, LM …
32
33
34
35
Subway Stations Bus Stations
36