Table 1.
POI categories.
Fig 1.
Partition of Beijing into 1km×1km blocks.
Fig 2.
Distribution of outflows at different time intervals.
A: 8:00-9:00; B: 13:00-14:00; C: 20:00-21:00.
Fig 3.
Distribution of inflows at different time intervals.
A: 8:00-9:00; B: 13:00-14:00; C: 20:00-21:00.
Fig 4.
The weights contributing to the traffic flows in 7:00-8:00 AM.
(a) Outflow; (b) Inflow.
Fig 5.
The weights contributing to the traffic flows in 13:00-14:00 PM.
(a) Outflow; (b) Inflow.
Fig 6.
The weights contributing to the traffic flows in 18:00-19:00 PM.
(a) Outflow; (b) Inflow.
Fig 7.
The weights contributing to traffic flows in 21:00-22:00 PM.
(a) Outflow; (b) Inflow.
Fig 8.
The average prediction accuracies at different time periods.
(a) The case of outflow; (b) The case of inflow.
Fig 9.
The average prediction accuracies for outflow under different κ in different time durations.
Fig 10.
The average prediction accuracies for inflow under different κ in different time durations.
Fig 11.
The average prediction accuracies for outflow under different hot degrees determined by β.
Fig 12.
The average prediction accuracies for inflow under different hot degrees determined by α.
Fig 13.
The average prediction accuracies against different time intervals during work days.
Fig 14.
The average prediction accuracies against different time intervals during weekends.
Table 2.
The average prediction accuracies (%) on different datasets in the case of outflow.
Table 3.
The average prediction accuracies (%) on different datasets in the case of inflow.
Table 4.
The average prediction accuracies (%) of outflows for different week days.
Table 5.
redThe average prediction accuracies (%) of inflows for different week days.
Fig 15.
The average prediction accuracies under different hot degrees with and without TF-IDF method.
Fig 16.
The average prediction accuracies against different time periods under different κ with different methods on hot regions with (α, β)=(1/2, 1/2).
(a) The case of outflow; (b) The case of inflow.
Fig 17.
The average prediction accuracies against different time periods under different κ with different methods on hot regions with (α, β)=(1/3, 1/3).
(a) The case of outflow; (b) The case of inflow.
Fig 18.
The average prediction accuracies against different time periods under different κ with different methods on hot regions with (α, β)=(1/4, 1/4).
(a) The case of outflow; (b) The case of inflow.
Fig 19.
The average prediction accuracies against different time periods under different κ with different methods on hot regions with (α, β)=(1/5, 1/5).
(a) The case of outflow; (b) The case of inflow.
Fig 20.
The average prediction accuracies against different time periods under different κ with different methods on hot regions with (α, β)=(1/6, 1/6).
(a) The case of outflow; (b) The case of inflow.
Fig 21.
The average prediction accuracies against different time periods under different κ with different models on hot regions with (α, β)=(1/2, 1/2).
(a) The case of outflow; (b) The case of inflow.
Fig 22.
The average prediction accuracies against different time periods under different κ with different models on hot regions with (α, β)=(1/3, 1/3).
(a) The case of outflow; (b) The case of inflow.
Fig 23.
The average prediction accuracies against different time periods under different κ with different models on hot regions with (α, β)=(1/4, 1/4).
(a) The case of outflow; (b) The case of inflow.
Fig 24.
The average prediction accuracies against different time periods under different κ with different models on hot regions with (α, β)=(1/5, 1/5).
(a) The case of outflow; (b) The case of inflow.
Fig 25.
The average prediction accuracies against different time periods under different κ with different models on hot regions with (α, β)=(1/6, 1/6).
(a) The case of outflow; (b) The case of inflow.
Table 6.
The average prediction accuracies (%) for outflow prediction using different methods.
Table 7.
The average prediction accuracies (%) for inflow prediction using different methods.