This article updates my ongoing compendium of TTC route performance statistics to include the first quarter of 2015.
Route_Performance_Summary_15Q1
The numbers reported by the TTC represent performance relative to the planned headways (time between vehicles), not to scheduled arrival times on routes:
The percent means that proportion of vehicles that were within +/- 3 minutes of the scheduled headway. More specifically that when each vehicle passes a ‘timing point’ it is compared to the vehicle in front of it that last crossed that same point. [From TTC Route Performance report]
In my version of the table, the TTC data are arranged in route number order with values for each quarter. Where there are blanks in the table, there was no data in the corresponding TTC quarterly report.
Also shown are the maximum, minimum, average and standard deviation values for each route’s statistics. This gives a sense of how much these values have moved around. A high standard deviation flags data that have widely varying values.
Of the 179 routes reported, 87 have higher ratings in 15Q1 than their averages for the nine quarters reported to date. To put it another way, about half of the routes did better than average for the first part of 2015, while about half did worse. If the extremely bad winter were a factor overall, one would expect a less balanced situation. On the streetcar lines, only three of eleven routes bettered their averages in 15Q1 (King, Lake Shore and St. Clair), but many of the differences are small with five of eleven falling within one standard deviation.
Headway reliability numbers are consistently bad on the 14x Downtown Express routes, and this implies that these infrequent services have a problem with running on time. What is not known is the measurements times and locations used to produce the stats for these routes and whether the service is at least on time where it collects passengers.
Similarly, headway measures for the 3xx Blue Night services are unimpressive, and what matters much more for these routes is on time performance and reliability of connections between routes, such as they exist.
The TTC claims that it will be introducing new “Journey Time Metrics” later in 2015, but there are as yet no details of what exactly these will measure. In parallel, there are moves to change the service reliability standards so that they look at routes end-to-end, not simply at their central points. (This was described in a presentation at the TTC board meeting in March 2015 (see p7 of TTC Modernization).
The TTC has yet to settle on a reporting mechanism that takes into account the difference in rider needs depending on the nature of a service. When a route is supposed to provide “frequent service”, the important point is that it be reliable. A 5 minute headway is not “frequent” if this actually means three buses every 15 minutes. When service is less frequent, then waiting times for off-schedule vehicles are a huge annoyance and on time performance is key. Short turns, of course, play havoc with both of these measures for riders who need a route beyond its common turnback points. Plans to measure the proportion of service that actually arrives at termini will highlight these problems.
Underlying all of this is the absence of a clear goal, a definition of what constitutes “good” transit service. Too often the goal has been to constrain cost increases and make the best of whatever resources the TTC has at hand.
Steve Munro 
I’m giving the current administration a year to see if they can pull out of the low point we are currently in. While it is easy to cut, cut, cut, it will take months to reverse the problems we are now in.
LikeLike
I don’t understand what the problem is for the TTC with regards to this … it is clearly up to the TTC Commissioners to decide what is good service … they should start with the subway lines and just say … good service is a subway every 5, or 3 minutes depending on time of day … and for each station track the number of times each day this doesn’t happen, when it doesn’t happen, and how much longer than 3 or 5 minutes a car takes more than 3 or 5 minutes to arrive … this is an insanely simple metric to track, and plotting it on a graph shows if there are time of day or other issues … aggregating it across multiple stations shows how quickly the system can recover from problems … and it records exactly what customers experience … you could add a few other metrics (number of trains running each hour, any alarms or known service issues).
For buses and streetcars <20 minutes we should move to the same system, and for anything over 20 minutes it's on time at each stop and you just sum the amount each vehicle misses by down to the second (with triple penalty if the bus leaves early) – you could also track the number of buses early/late to each stop per day.
These types of things are as you know are insanely easy to calculate with the Nextbus information and Amazon cloud, there is almost no excuse at this point that this information isn't being provided on a daily basis, on a monthly aggregate basis and a system basis to the public.
That the commissioners haven't got into this is a dereliction of duty … that they seem to be oblivious that this information could be generated for them, and that they could use it to determine service levels and improve things with almost no cost. There should be a subcommittee on reliability or something.
From that report, I don't understand why they are moving to a concept of "end of line" and "midpoint" analysis … this is ridiculous … most of your customers don't get on at the end-of-line or midpoint … figure out the numbers for EVERY STOP … if you can figure out the numbers for one point, you can do it for EVERY POINT! If I'm at Dundas and Parliament, I honestly don't care when the Parliament bus left Castle Frank Station … I care when it will show up at Dundas and Parliament.
Steve: Yes, but irregular departures at terminals (not to mention from major points enroute) are a measure of how well (or not) the line is being managed. Also, there are a lot of riders who use the ends of various transit routes, or would if the service actually showed up.
LikeLike
TTC measure against headway (rather than actual schedule) is useless for low-frequency routes. If a half-hourly service always runs 5 minutes early, then it will count as 100% on-time, but means a delay of 25 minutes for 100% of passengers starting their trip on that route
LikeLike
Yes, and if I measure service at the end of line, not mid point, I am also capturing the use of short turns. I would agree that the measurement should include mid points, however, there is a limit to what you want to report on, and which points make the best proxy for the entire route. If a bus goes to the end of the line, it is reasonable to suggest that it went past all the intermediate points, however having said it went past all the intermediate points on time, does not suggest that it actually serviced the end of the line.
LikeLike
Steve: When I used to take the 141 Mount Pleasant Express (many years ago, but I suspect the same problem applies) the morning service was accurate almost to the second. However, the evening service was an unmitigated disaster, with buses arriving after the next bus was due. Delays of up to 30 or 40 minutes after the supposed arrival time were not uncommon. The problem was before the pick up point on Adelaide at Bay. The buses insisted on using only the curb lane west of Bay – which is subject to extreme gridlock due to cars turning right onto Bay. The next lane over would move from a point where bus first comes in sight to the stop in about 5 minutes or less. The bus used to take 20 minutes or more to cover the same distance. The evening service was extremely lightly used compared to mornings.
LikeLike