Q70-SBS Bus LaGuardia Airport

### Data Exercise – Determine Q70 SBS Travel Times (12/10)

Data: MTA bus data (all-records.csv file is provided for 12/10)
#### Task1: Calculate Q70 SBS travel times between the following Stop IDs:
    Stop ID    Name    Direction
    553332    ROOSEVELT AV/61 ST    NB
    551576    74 ST/ROOSEVELT AV STATION    NB
    804110    LaGUARDIA INTERNAL RDWY/TERMINAL B    NB
    503875    LGA/TERMINAL D    SB
    503848    LGA/TERMINAL C    SB
    505192    ROOSEVELT AV/74 ST    SB
    982075    61 ST/ROOSEVELT AV    SB

#### Task2: Visualize travel times and bus delays. 

My first step is to understand the problem by visualizing the bus route and the travel pattern between stations:

(You can select to see the complete route or the route divided by sections.)

Then, I will look at the time that the bus spend travelling between each station following the bus route and travel pattern.

Note: the data provided shows the buses following different patterns, the selected one is the most consistent, the other patterns could be because of errors in the data. This needs further analysis (not covered here).

After getting the travel times between stations in minutes I want to visualize the distributions in the different sections of the bus route:

It is clear that the are some outliers that need to be reviewed, travel times with 0 seconds or more than 120 were filtered. Still, the sections 505192-982075 and 503875-503848 show some outliers.

Below we can explore how the travel-time changes during different day times.

(You can select the different sections to observe them individually.)

Mean Times between stations:
MTA_553332-MTA_551576: 7.72 minutes
MTA_551576-MTA_804110: 9.82 minutes
MTA_804110-MTA_503875: 4.17 minutes
MTA_503875-MTA_503848: 2.65 minutes
MTA_503848-MTA_505192: 7.51 minutes
MTA_505192-MTA_982075: 18.1 minutes
MTA_982075-MTA_553332: 2.17 minutes

 

Looking at the plots we can see that the are some outliers affecting the travel times, with boxplot would be easy to indentify outliers:

Given the noise in the data the median provided by the Boxplot (hover over the box) is a better statistics estimate the travel time between each stations.

It is important to notice that the data provided did not contain scheduled time (all values empty), so it is not possible to obtain the delays as there is no scheduled time for reference. 

Another important exercise consists in observing where the buses spend most of their time. As every measurement from the GPS is recorded every minute, we can observe that the density of the GPS coordinates is concentrated in the south stations, especially:

  •   553332    ROOSEVELT AV/61 ST    NB
  •   982075    61 ST/ROOSEVELT AV    SB

If some improvement is required, looking at the traffics dynamics close to these stations would be a good starting point.

Bellow the map with the density of GPS coordinates from all the buses during December 10th.

Fernando MelchorComment