How have usage patterns of the late night MBTA and Taxi services changed between 2013 and 2014? We can explore this question using the interactive charts on this webpage.
These charts show the change between 2013 and 2014, on comparable weekend nights, by looking at the number of transactions. We have data for Feb 28th up to Jun 7th of this year, omitting the third Friday in April because of highly unusual circumstances occurring on Apr 19th, 2013.
Clicking on one chart will update the other two, causing them to focus on the desired time and/or place only. For example, you can look at May 3rd, 2014 to see that there was an increase of 6,357 transactions compared to the previous year.
In addition, you can switch between MBTA view and Taxi view with the big slider button on the top left of the page, and you can clear filters by clicking the "X" mark next to them on the upper right. You can always reset the charts back to their initial settings by clicking this link: reset.
At first glance, you should be able to see some striking differences in the MBTA dataset, caused by the introduction of late-night service hours. Before the last weekend in March, ridership had been trending slightly downward compared to the previous year. Starting on March 28th, there is a sharp increase, as you can see on the calendar, where the dates shift from a light orange color to a distinct blue. The difference can also be viewed on the bar chart that breaks down the 15-minute time periods. The orange bars start to decrease and trail off by 1 a.m., but the blue bars level off and stay stronger until around 2:30 a.m. in the morning. This represents the fact that MBTA usage past 1 a.m. is significantly higher in 2014 compared to 2013, as we expect.
Let's take a look at the very first day of late night service, March 28th. This day saw an increase in late night ridership of 6,620 compared to the previous year. You will also see a pattern that recurs frequently, where ridership earlier in the evening declines slightly, but is more than overwhelmed by ridership late at night. This is not due to baseball: it was still spring training in both cases. One hypothesis might be that people are staying out later. Another might be that there is a secular decline in ridership that is being masked by the additional late night service. More analysis is required.
Although that particular example didn't show it, you can often see the effects of baseball games at Fenway Park in the data. Check out April 4th focused on Kenmore Square. The blue bars dominate the time chart, showing massive peaks of ridership from fans leaving the game, over the course of the evening. The difference is that the Red Sox hosted the Brewers on this weekend in 2014, whereas the Sox were in Toronto the previous year. The influx of riders from Fenway Park is so strong that it completely dominates the data when you are focused on Kenmore Square, like in this example. The patterns of colors in the Kenmore Square-focused calendar are driven by the Sox evening home game schedule.
Let's shift back to the late-late night crowd and take a look at how things changed in the 15 minute period beginning when the bars close at 2 a.m. Again, it should come as no surprise to see sharp increases in MBTA ridership at 2 a.m. Check out the list of routes and stations that have seen the increases: Park Street has seen over 3,000 new boardings between 2 a.m. and 2:15 a.m. in the provided data alone. Where did these riders come from? Presumably, some of them used to take taxis.
Let's flip the switch and take a look at Taxi data for the same 15 minute period of time. Immediately we can see a lot more orange, meaning that there has been a decline in taxi usage between 2013 and 2014. But does this mean that former taxi riders are switching to late night MBTA in droves? Well, not exactly. What we can see is that there seems to be a year-long decline in taxi transactions: even before March 28th, taxi usage was going down about as sharply as it does after. In addition, based on the time chart, there seems to be a decline in taxi usage prior to midnight that is just as big, if not more so, as after midnight.
Not all areas have seen a decline in taxi usage at 2 a.m. In particular, trips starting in parts of downtown Boston, as well as Dorchester, East Boston and Jamaica Plain/Roxbury have increased in the past year compared to 2013. But when comparing those to the starting ZIP codes that have decreased, it's not really clear what the difference may be. The decline in late night taxi ridership seems to be occuring primarily in areas served by the Green Line, as well as parts of Cambridge and South Boston. There is a particularly large drop found in 02215 (Kenmore/Fenway/BU), 02114 (West End/Bulfinch Triangle) and 02134 (Allston), which happen all to be areas with significant nightlife. On the other hand, the increases in taxi usage are found in places like 02108 (Beacon Hill/DTX) and 02110 (Financial District/Greenway), which are somewhat quieter but still have significant late night activity in some quarters.
Enough about taxis and downtown bars. I'm curious about the new late night service hours and their effect on buses in neighborhoods not well-served by rapid transit. I think there was a reasonable worry that the new service hours would be seen and known about mostly in the subway corridors, leaving the bus riders "out in the cold." Let's examine late night ridership of key bus routes 15, 22, 23 and 28, which are all important routes branching out southwards from Dudley. Interestingly enough, although late night service officially started on March 28th, it isn't until April 11th that late night ridership on these bus routes started picking up. That might be a result of news traveling more slowly, or perhaps some initial skepticism about the service -- who wants to wait on the corner for a bus that may never come?
Perhaps, once the late night service was seen to be a real thing, it started to catch on. Very strongly, it appears, in fact. So strongly, that bus ridership has been boosted at all times between 22:00 at night and 3:00 in the morning. This can be seen, for example, by comparing these buses on March 29th to June 7th and having a completely different picture emerge in just two months. On the first weekend of late night service, you see year-over-year declines in ridership compared to 2013. Yet, a few weeks later and it's the complete opposite: now there are strong increases in year-over-year ridership compared to 2013. At all hours: not just in the new service hours, but also before 1 a.m.
I could spend a long time playing with this data. I just want to mention one of the more surprising things I found by accident, which is the curious popularity of the major park-n-ride stations late at night. Ordinarily, you would expect most people who use those stations to be leaving the city at these hours, and not registering as a transaction in the MBTA automated fare collection system. Yet, when you plug in Oak Grove, Braintree, Quincy Adams, Riverside, Alewife, Wonderland, Wellington you end up seeing a fairly clear difference before and after March 28th, and strong late night ridership shown in the time chart. My guess is that these riders are either people being dropped off for a late night trip back to the city, or possibly they are very early hours workers taking advantage of the MBTA and the relative emptiness of the park-n-ride lots at this hour.
I won't go on forever, and will leave the rest of the exploring to the reader, for now. Please enjoy clicking around and trying out different settings. One thing I will point out is that all of the machinery behind this can easily be extended with any additional data from the intervening weeks that the MBTA/MassDOT decides to release. This site was created using Google Spreadsheets, Google Charts, GitHub and Javascript.