ON-TIME PERFORMANCE 6/12/04 THROUGH 7/11/04

Results for the most recent period reviewed are:

TOTAL TRAINS = 1020

ON TIME = 181 (18%)

LATE = 839 (82%)

AVERAGE DELAY = 1' 25"

Only schedulled arrival times were used. No "grace" period was allowed.

Of the late trains:

261 (31%) were 29" or less late;
220 (26%) were 30" to 59" late;
177 (21%) were 1' to 2' late;
178 (21%) were more than 2' late; and
3 (<1%) had no arrival time shown.

BY ROUTE, WITH ON-TIME % AND AVERAGE DELAY:

CAPITOL LIMITED
#29 - 7% - 2' 38" #30 - 0% - 3' 49"

THREE RIVERS
#41 - 20% - 2' 44" #40 - 7% - 2' 28"

PENNSYLVANIAN
(W) - 23% - 48" (E) - 0% - 40"

LAKE SHORE LIMITED
#49 - 3% - 2' 21" #48 - 3% - 2' 26"

LAKE SHORE LIMITED (BOSTON / ALBANY)
#449 - 17% - 1' 31" #448 - 0% - 2' 08"

CITY OF NEW ORLEANS
#59 - 20% - 1' 02" #58 - 27% - 1' 15"

CRESCENT
#19 - 13% - 1' 30" #20 - 57% - 1' 11"

DETROIT CORRIDOR
(W) - 19% - 46" (E) - 19% - 36"

ST. LOUIS CORRIDOR
(W) - 23% - 45" (E) - 16% - 2' 05"

SYRACUSE CORRIDOR
(W) - 33% - 36" (E) - 11% - 1' 07"


This was another disappointing period as there was no turnaround from the low figures of the prior period.

The only services to show improvement in both OTP and Avg. Delay were the westbound Pennsylvanian, and both the westbound St. Louis and Syracuse Corridors.

There were 94 trains that departed their initial terminals (BOS, NYP, CHI, NFL, PNT, and NOL) so late that an on-time arrival was not possible. 53 of these were delayed more than 1', up to the longest of 4' 15".

Of the trains arriving more than 2' late, the Lake Shore Limited accounted for 45, the Capitol Limited for 37, the Three Rivers for 29, the St. Louis Corridor for 20, the City of New Orleans for 10, and the Crescent for 9.
49 of these trains were delayed more than 4', 9 of which delays were between 8' and 11'.

The only longer distance train to operate at 50% or better was the eastbound Crescent, and the only short-distance train to do so was #283 on the Syracuse Corridor.

While some delays are not the responsibilityof either Amtrak or the host railroads (e.g.: #64 on the Syracuse Corridor clearing customs en route from Toronto), the overwhelming majority are. The failure to eliminate the avoidable delays continues to negatively impact the services, resulting in the unacceptably low OTP figures.

While some delays are not the responsibility of either Amtrak or the host railroads (e.g.: #64 on the Syracuse Corridor clearing customs en route from Toronto), the overwhelming majority are

Are what?

And if we go back to the "either/or" as mentioned, what are the percentages thereof?

Chieftan,

They are the responsibility of Amtrak or the host railroads.

As to a breakdown of the responsibility for delays to try for percentages, there is insufficient data to separate those of Amtrak and those of the host carriers.

shlustig wrote:As to a breakdown of the responsibility for delays to try for percentages, there is insufficient data to separate those of Amtrak and those of the host carriers.

Absolutely false. In fact, we went through all this last month.

This breakdown is available in Amtrak's Monthly Performance Reports. Here's a direct quote from page 106 of April's report:

Amtrak wrote:On the six major host railroads, Host-Railroad Responsible delays accounted for 70 percent of all delay minutes in April . . . . Freight Train delays, Slow Orders, and Signal Delays represented 30, 21 and 11 percent of all minutes, respectively . . . . Amtrak- Responsible Delays accounted for 15 percent of minutes; and Passenger and Commuter Train Interference accounted for 10 percent

Got that? 70% of the delays over the hosted part of the system can be traced back to the host railroads.

But wait, it gets better: Amtrak compiles and releases delay information for EVERY hosted train over the ENTIRE non-Amtrak portion of the route. Not just a relative handful of trains measured at the endpoints only.

mattfels wrote:

shlustig wrote:As to a breakdown of the responsibility for delays to try for percentages, there is insufficient data to separate those of Amtrak and those of the host carriers.

Absolutely false. In fact, we went through all this last month.

This breakdown is available in Amtrak's Monthly Performance Reports. Here's a direct quote from page 106 of April's report:

Amtrak wrote:On the six major host railroads, Host-Railroad Responsible delays accounted for 70 percent of all delay minutes in April . . . . Freight Train delays, Slow Orders, and Signal Delays represented 30, 21 and 11 percent of all minutes, respectively . . . . Amtrak- Responsible Delays accounted for 15 percent of minutes; and Passenger and Commuter Train Interference accounted for 10 percent

Got that? 70% of the delays over the hosted part of the system can be traced back to the host railroads.

But wait, it gets better: Amtrak compiles and releases delay information for EVERY hosted train over the ENTIRE non-Amtrak portion of the route. Not just a relative handful of trains measured at the endpoints only.

As far as publicly-accessible data goes, one can find out the percentage of delay-minutes attributable to different sources. However, that doesn't equate to total number of trains delayed. For example, if you take a 16-hour late Sunset Limited (assuming all delays are host-related) and 16 one-hour delays that are Amtrak's fault, then you get a total of 50% for each.

Given the Sunset Limited's tendency to run hours and hours (and hours) late on the Union Pacific (and the CSX doesn't necessarily make things any better), the numbers are a bit skewed towards the host railroads. And the data that shlustig collected (presumably amtrak.com reports) did not specify cause of delays.

The Sunset Limited doesn't appear in the summary.

As far as publicly-accessible data goes, one can find out the percentage of delay-minutes attributable to different sources. However, that doesn't equate to total number of trains delayed.

And neither does this summary. It doesn't cover the whole system, and the trains it does cover are measured only at the endpoints. If your destination or embarkation point is, say, Cleveland, this summary is useless. Good for bashing, and very little else.

mattfels wrote:The Sunset Limited doesn't appear in the summary.

As far as publicly-accessible data goes, one can find out the percentage of delay-minutes attributable to different sources. However, that doesn't equate to total number of trains delayed.

And neither does this summary. It doesn't cover the whole system, and the trains it does cover are measured only at the endpoints. If your destination or embarkation point is, say, Cleveland, this summary is useless. Good for bashing, and very little else.

Amtrak's documents don't mention midpoints either, nor do they indicate the extent of the delays (i.e. average delays per train). This summary at least provides the latter. On-time percentage by itself means nothing. Two trains could have the same percentage, but one could run hours late on the trips when its late, while the other might miss its on-time arrival by 15-30 minutes.

Won't find anything like that in this summary either. The "average" delay of the delayed trains still doesn't tell you anything useful. If you want to know the chance that your train will be significantly late, shlustig's "no grace period" policy muddies things up considerably. Example: 8 trains on a given route arrive their endpoint 15 minutes late; a ninth arrives 4 hours late. Shlustig will tell you only that 9 of those trains were late, with an average delay of 40 minutes. Even if you're detraining at the endpoint, does this tell you anything worth knowing? Not really.

Of course, if your train happens to be the Empire Builder, you're completely outta luck. Why? Didn't make the cut.

One of the few things I really understood from my grad school...

A statistic that has questionable "face" validity is worse than no statistic at all.

I'm with Matt here. Give us a hyperlink, and we can do the math for ourselves. Mr Lustig's effort, while energetic, is partial at best and, imnsho, has questionable face validity.

John

John_Perkowski wrote:Give us a hyperlink, and we can do the math for ourselves.

http://tickets.amtrak.com/Amtrak/trains ... lCall=true

Have fun.