Academia.eduAcademia.edu

Figure 2 – uploaded by Marco Veloso

See full PDFdownloadDownload figure
Fig. 2. Taxi service variation according to the hours of day (a) and days of week (b). The following figure shows the temporal variation of the taxi services. As expected, it gradually increases in the morning reaches the maximum between 11 a.m. and 1 p.m, and slowly drops down in the afternoon. In the working days, there are more taxi services than in the weekends with the maximum number of the services observed on Monday.

Figure 2 Taxi service variation according to the hours of day (a) and days of week (b). The following figure shows the temporal variation of the taxi services. As expected, it gradually increases in the morning reaches the maximum between 11 a.m. and 1 p.m, and slowly drops down in the afternoon. In the working days, there are more taxi services than in the weekends with the maximum number of the services observed on Monday.

Related Figures (10)

Fig. 1. Taxi pick-up (a) and drop-off locations (b) distribution. The overall taxi service distribution in Lisbon can be seen in Fig. 1 where some major locations such as the city downtown (A), the airport (B), and the train station (C) are identified.
Fig. 1. Taxi pick-up (a) and drop-off locations (b) distribution. The overall taxi service distribution in Lisbon can be seen in Fig. 1 where some major locations such as the city downtown (A), the airport (B), and the train station (C) are identified.
Fig. 3. Distribution of trips distances. which seems unrealistic. Therefore, in addition, we discarded these small trips from our analysis.
Fig. 3. Distribution of trips distances. which seems unrealistic. Therefore, in addition, we discarded these small trips from our analysis.
Fig. 4. Distribution of trips duration. The trip duration distribution is represented by the following figure. Similar to the trip distance distribution, we discarded trips that are less than one minute and longer than two hours. After the data cleaning process, we retained 177,169 trips and 217 distinct taxis.
Fig. 4. Distribution of trips duration. The trip duration distribution is represented by the following figure. Similar to the trip distance distribution, we discarded trips that are less than one minute and longer than two hours. After the data cleaning process, we retained 177,169 trips and 217 distinct taxis.
Fig. 5. Distribution of trips distance (solid line) and fitted Gamma distribution (dashed line).
Fig. 5. Distribution of trips distance (solid line) and fitted Gamma distribution (dashed line).
Fig. 6. Trip distance distribution without the first interval of the dataset (solid line) and fitted exponential distribution with 4 = 0.26 (dashed line). If the first interval is removed, we can represent it with an exponential distribution, exp(A) with 4 = 0.26 (Fig. 6). The same phenomenon (exponential distribution) can also be observed for the trip duration and the trip income.
Fig. 6. Trip distance distribution without the first interval of the dataset (solid line) and fitted exponential distribution with 4 = 0.26 (dashed line). If the first interval is removed, we can represent it with an exponential distribution, exp(A) with 4 = 0.26 (Fig. 6). The same phenomenon (exponential distribution) can also be observed for the trip duration and the trip income.
Table 1. Statistics for the top five taxi drivers, considering total number of trips and income. Table 2. Statistics for the top five taxi drivers, considering the number of trips and income from the airport.
Table 1. Statistics for the top five taxi drivers, considering total number of trips and income. Table 2. Statistics for the top five taxi drivers, considering the number of trips and income from the airport.
We find that the majority of the taxi drivers appear to be using combined strategies — mainly driving around the city and in certain time periods staying at a fixed location. This phenomenon can be observed in the Fig. 7 where there is an increase of trips from the airport between 6 a.m. and 8 a.m. and elsewhere in other time slots. On the other hand, in the same period there is a significant decrease of trips from the remaining locations.
We find that the majority of the taxi drivers appear to be using combined strategies — mainly driving around the city and in certain time periods staying at a fixed location. This phenomenon can be observed in the Fig. 7 where there is an increase of trips from the airport between 6 a.m. and 8 a.m. and elsewhere in other time slots. On the other hand, in the same period there is a significant decrease of trips from the remaining locations.
Fig. 7. Variation of number of trips aggregated by hour. Left: pick-up location from the airport. Right: pick-up from other locations. providing statistical measures of carried load (Erlang) with one-hour period. In Fig. 8. we can observe that immediately after the taxi rush hour there is an increase of the GSM network usage in the airport, while the mobile phone activities in othet locations of the city begins to arise three hours later. This is an indication of the possible taxi passengers at airport.
Fig. 7. Variation of number of trips aggregated by hour. Left: pick-up location from the airport. Right: pick-up from other locations. providing statistical measures of carried load (Erlang) with one-hour period. In Fig. 8. we can observe that immediately after the taxi rush hour there is an increase of the GSM network usage in the airport, while the mobile phone activities in othet locations of the city begins to arise three hours later. This is an indication of the possible taxi passengers at airport.
Fig. 8. GSM network usage aggregated by hour.
Fig. 8. GSM network usage aggregated by hour.
Fig. 9. Distribution of (a) pick-up locations with destination the airport and (b) drop-off locations from the airport.
Fig. 9. Distribution of (a) pick-up locations with destination the airport and (b) drop-off locations from the airport.
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved