A Time Series Analysis of Transit Ridership
A Time Series Analysis of Transit Ridership
Presently over 500 federally assisted transit agencies provide service in urbanized areas of the United States. These transit services consist of heavy rail, light rail, commuter rail, trolley bus, and buses (as well as aerial tramways, automated guideways, cable cars, ferryboats, inclined planes, and monorails). Federal law requires these transit agencies, with a few exceptions for the very small agencies, to report their financial and operational activities to the Federal Transit Administration.
Transit ridership is a count of unlinked trips, which is defined to be a trip on one transit vehicle regardless of the type of fare paid or transfer presented. The American Public Transportation Association (APTA) estimates that the number of people riding transit on an average weekday is 45% of the number of unlinked transit passenger trips.
Figure 1 - U.S. Transit Ridership
Figure 1 provides a graph of the monthly number of unlinked trips on U.S. public transit. As is true of other monthly transportation indicators, transit ridership is highly seasonal. Behind the seasonal fluctuations is a long-term pattern that appears to be increasing during the late 1990s. Well statistically decompose the data into the seasonal, trend and irregular components to verify this observation.
Figure 2 - Seasonal Average (1992-2001)
The analysis of the seasonal behavior of the data indicated that the monthly variations varied slightly over time. However, no particular pattern in that variation was apparent (i.e., no particular month showed a consistent decrease or increase over the past nine years). Figure 2 provides a graph of the monthly variations averaged over the past nine years, in order to compensate for the variations in the monthly deviations over time.
The histogram in Figure 2 indicates that October is the month of particularly high ridership, whereas the summer months and February tend to be lower than average.
By removing the seasonal and irregular components of the time series, we can now observe the long-term trend underlying the transit ridership. Figure 3 displays a plot of the actual ridership along with a plot of this underlying trend. (Please note that the scale for the y axis has been modified to show greater detail in the data.)
Figure 3 - U.S. Transit Ridership (Actual and Underlying Trend)
The trend in Figure 3 reveals that an upward growth in ridership began in early 1996. This positive trend continues through to the present. The positive slope of this long term trend represents a increase of over 900,000 trips per month.
In order to specify what mode within transit ridership was driving the trend from 1996 through to the present, we created a plot of the annual ridership by mode (see Figure 4). The graph indicates that both bus (which presently constitutes 60% on the trips) and heavy rail (which presently comprises 28% of the trips) showed strong positive growth from 1996 onward.
Figure 4 - U.S. Transit Ridership (Annual Modal Split)
