Analysis of Route 39 Finch East: Part II — Monthly Views

In this post, I will review the behaviour of the Finch East route over the month of December 2006.  As I mentioned in Part I, the schedules have changed a bit since then, most notably in the extension of a frequent service to Neilson Road in the PM peak period.

First let’s look at where all the buses went outbound, and how well-spaced the service was.  This chart shows the destination of all buses eastbound from Victoria Park.  The horizontal spacing measures the headway at that location, and the vertical axis measures how far the bus actually went.  The ongoing problem with service reliability to Neilson Road is quite evident on all weekdays except for those when special schedules operated.

Destinations eastbound from Victoria Park

In this chart, any short turns at Seneca College have already been filtered out as they are west of the reference location.

On every weekday, the clustering of vehicles is evident and this even show up at times during periods of the supposedly 12 minute headway to Neilson Road. There is a variation in what I can only call the “texture” of the service from day to day, but the overall pattern is clear. When service is very frequent, close management is not generally needed as long as the overall scheme is maintained. However, as is evident, there are gaps and when these are on the less-frequently served part of the route, they can be quite large.

Service to Neilson may not actually be as bad as it looks for some riders who can use the 139 Finch East service merging in eastbound at Don Mills. However, the scheduled headways in December 2006 didn’t quite match (one was 12, one was 14 minutes), and so even a perfectly on-time operation would produce periods of moderately well-spaced service followed by pairs of buses.

It’s the lack of reliability that really frosts riders, and that doesn’t seem to be a goal judging by the variation in service from day to day.

On both Christmas and Boxing day, pairs of buses can be seen travelling east from Victoria Park now and then.

A special note about December 29: the wide gap is caused by missing CIS data for the period between about 12:20 and 2:00 pm. This is common to data for all routes that I have on this date. December 22 and 29 have special schedules with the PM peak service starting at noon.

Headway Charts

Each set of headway charts contains seven pages.  The first four show data for the weekdays in each week with trend lines for each day.  The fifth shows the data for all weekdays to give a sense of the overall behaviour, how tightly bunched (i.e. consistent) the data are, and whether major swings in values affect all days or only a few.  The sixth and seventh contain Saturdays and Sundays.  Boxing Day is included with Saturdays, and Christmas is included with Sundays.

Headways westbound at Tapscott
Headways westbound at Victoria Park
Headways westbound at Willowdale

At Tapscott, the variation in headways inbound from Neilson, especially in the PM peak, is quite striking.  Some gaps well above the 12 minute schedule are visible in weeks one through three.  Week four settles down with many, but not all, of the data points lying within the target six minute “on time” band.

Weekend headways at Tapscott are fairly well behaved although the number of values close to zero corresponds with the bunching we see on the Destination Chart above.

By the time we reach Victoria Park, all of the short-turn services except Seneca College have merged into the line.  The data is clustered roughly where one would expect for the combined headway although there are outliers.  Note how the headways become more dispersed in the evening when less frequent service is more affected by poor headway adherence.

The pattern for weekends is comparable with the greater scatter applying to periods of overall less frequent service.

By the time we reach Willowdale, weekday headways are generally in a band from 0 to 6 minutes with some outliers.  Weekend headways show the same pattern as elsewhere on the line with more scatter overall and particularly in the early and late parts of the day when scheduled headways are wider.  The wide range in late evening headways is particularly troubling.

Headways eastbound at Willowdale
Headways eastbound at Victoria Park
Headways eastbound at McCowan
Headways eastbound at Tapscott

Eastbound at Willowdale, the headways are more tightly bunched around the target value, but we see the same pattern with dispersion of values late in the evening.  Considering that this is quite close to Finch Station, this can only be explained by operator behaviour, not by some mysterious source of congestion.

By the time we reach Victoria Park, the headways are starting to spread out as pairs begin to form.  This continues to McCowan, the major turnback point on this route.

Service eastbound at Tapscott during the PM peak shows the same pattern we saw westbound with major gaps during the weeks before Christmas.   Week four is better behaved.

Link Times

The following charts have the same seven-page layout as the headway charts above, but they track the travel times between points on the route. The variation (or lack of it) shows the degree to which congestion and/or loading delays play a regular or occasional role in route operations.

Link times westbound from Victoria Park to Seneca College
Link times westbound from Seneca College to Leslie
Link times eastbound from Leslie to Seneca College
Link times eastbound from Seneca College to Victoria Park

I have included only two of the segments of the route because overall these charts do not show very much. There is some congestion visible on some days in the trip through the region where DVP spillover will have an effect, but it’s not very big in these stats. (Comments in the thread for Part I suggest that this sort of congestion is an ongoing issue on the route.)

Overall, the times are generally uniform, with a small number of peak period exceptions, westbound from Victoria Park to Seneca College, and from Seneca to Leslie. Similarly, eastbound times show some peaking on individual days, but not major changes in running times such as those we see on downtown streetcar routes.

However, when we look at the route over a longer part of the trip, the picture is quite different.

Link times westbound from Tapscott to Willowdale
Link times westbound from McCowan to Willowdale
Link times eastbound from Willowdale to Tapscott
Link times eastbound from Willowdale to McCowan

The Tapscott charts pick up only the service running through to Neilson Road, while the McCowan charts contains much of the route’s service. The westbound service show some peak period effects especially in week one. Weekend service has no peaks indicating that there is no location with shopping related congestion such as we saw on Dufferin.

However, it is worth noting the variation in times over a span of 10 minutes for some trips on weekends. Indeed, some weekday periods have more tightly clustered times. This may show both variations in operator practice (some running as fast as possible) as well as the availability of enough running time for breaks along the route.

The times westbound from McCowan to Willowdale show a more pronounced peak effect because the congested part of the route contributes more to the total.

Eastbound times are much more spread out than the westbound data indicating greater problems with congestion in that direction.

(As a general note, there will be some dispersion in the McCowan charts because of the mixture of express and local trips, but this should make the same contribution every day.)

The interesting part about these charts when you look at every segment of the route (I have not published them all here) is that no individual segment stands out as a major source of congestion. This is much different from Queen where one can point to areas with running times that double or worse during certain periods, and others where they are completely flat. What appears to be happening on Finch East is that much of the route contributes more or less equally to the overall variation in trip times.

This means that a “quick fix” targeting a few locations will not produce a big difference on the route.

Terminal Times

Finally we come to the time spent at and near terminals.  As I have written before, accurate resolution of terminal operations is difficult with CIS data, but we can use the round trip times to a nearby point as a surrogate.  For example, the time taken to go from Willowdale west to Finch Station and return will largely be determined by the time spend entering, serving and leaving the station.  The time east of Tapscott (which I used as an east end reference point) out to the Neilson and Crow Tail loop and back is a rough indication of the variation in layover time available at the outer end of the line.

Round trip times from Willowdale to Finch Station
Round trip times from Tapscott to Neilson

At Finch Station, the round trip from Willowdale lies roughly in a band around ten minutes with a spread of +/- five for all weekdays.  This is remarkably consistent.  On Saturdays, the value is roughly the same, although it drops off late at night when schedules are tigher and the opportunity for layovers is reduced.  Sundays show the widest range of values probably due to generous running times.

At Neilson, the times are more tightly clustered, but with greater dispersion in the evening.  This suggests that, during the day at least, buses do not sit for long at the east end of the line but return as soon as they can.  As at Finch Station, weekend values are more dispersed with some evening runs having a quite leisurely trip from Tapscott to Neilson and return.

In the next and last installment for Finch East, I will present details of a few days of interest.

6 thoughts on “Analysis of Route 39 Finch East: Part II — Monthly Views

  1. re: ‘Destination Eastbound from Vic Park” chart

    Steve, either there’s NO service at all or something a bit wack with the chart! No lines plotted on it 😦

    Steve: I just looked at the chart as a download from my site and it looks fine with lots and lots of lines. Is anyone else having trouble with this?


  2. Finch East eastbound is a pain in the neck with buses short turning in the afternoon peak. I used to take the 139E from Don Mills station to Neilson & Finch for a while as its never crowded but gave up on it due to too many late buses and short turns at McCowan or Middlefield in the afternoon peak around 5:45-6:00 pm. Now I take the 39E from Finch Station to Neilson which is usually packed (but has an advantage with air conditioned buses) and still have a reoccurring problem with short turns in the afternoon rush. There have been times when there is a short turn three or four days in a row and traffic is horrible along Finch much of the way from Yonge st. In the past I have gone through Kennedy Station and the LRT in the afternoon rush but found that to be too crowded and have too many delays like no winter service. The trip between Neilson & Finch in Malvern and downtown on a good day takes 1.5 hours and on a bad day takes 2 to 2.5 hours.


  3. I see the chart fine. I wonder if bbpsi is looking for your multi-coloured lines used in the streetcar service analysis; as opposed the black/grey bar graph form.

    Steve: Actually this chart was deliberately generated with only black. The bars are black with no borders. Sounds like a rendering problem. I will have friends who run OSX have a look at it to see what they see.


  4. I’m viewing the PDF on a Mac in Preview. (Mac OS X built-in PDF viewer.)

    It’s supposed to be able to properly display all standard PDFs, but its possible there is a compatibility issue there..


  5. I’ve been on this route in the fringes of off-peak (IE the last few buses on a sunday night) and the one thing I’ve always noticed is how fast they go. It strikes me that adding some terminal time might be useful in some instances (it sounds like the afternoon peak is worse then the morning, and nielson is worse then finch station. so add it out there and then.)


  6. Figured it out. Its the weight of the lines. Preview on OS X chooses to not render really thin lines at all unless you are zoomed in closely.

    In other words, if you zoom in a few steps on the graph they become visible, but if you leave the chart at the standard screen-filling size they’re invisible.


Comments are closed.