Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

< Back to Article

Table 1.

Summary statistics of one-week of smart-card data (metro trips only).

More »

Table 1 Expand

Fig 1.

Variability of temporal patterns of trip starting times on the London underground.

More »

Fig 1 Expand

Fig 2.

Variability as a function of the time intervals for London, Singapore, and Beijing.

Note: The negative linear relations show that the variability declines with increasing temporal scale but the variations in the variability (the variance of the covariance) which relate to the stations also increase for all three cities.

More »

Fig 2 Expand

Fig 3.

Distribution of the variability of trip starting time at London underground stations.

Note: Less variance as the temporal scale gets finer while log normality is consistent over all scales.

More »

Fig 3 Expand

Fig 4.

Variability of regularity in the trip matrix over time.

Note: Each box plot shows the variability of 400 stations over time measured at different temporal scales. Overall, eight subplots give a similar trend where lower variability appears during peak hours (around 9 am in the morning and 6pm in the evening). More details can be captured as differences of variability between each time unit are magnified as we decrease the temporal scale from 12h to 4 minutes.

More »

Fig 4 Expand

Fig 5.

Densities of variability for different time intervals for 400 London underground stations.

Note: Less variance exists at a certain time slot, while log normality is consistent over all time slots at all scales. More details about the differences show up when the temporal scale decreases.

More »

Fig 5 Expand

Fig 6.

Comparative analysis of variability in temporal patterns.

More »

Fig 6 Expand

Fig 7.

Comparative analysis of predictable trips origin and destinations.

Note: A critical point exists at 15 minutes in all three cities as a universal pattern, which implies how closely we can predict future event.

More »

Fig 7 Expand

Fig 8.

Combined ranking of when and where people travel.

Note: Stations are ranked by CVarcutoff = 0.1 and measured using temporal scales = 4, 15, 30, 60 and 180 (minutes). Stations which fulfill conditions at 4 minute temporal scales get the highest rank as 5 and so on. The geographic mapping is color coded by a combined score.

More »

Fig 8 Expand