Analysis of Commuter Blog

Tue 8 Jan 2008, 13:30

I have just completed an analysis of the Commuter Blog postings for November and December. Once again, all the figures refer to weekday services only. The figures are as follows:

November:

195 trains were surveyed (28.6% of all weekday services for the period).

Nine trains were cancelled, giving a reliability rating of 95.4% for those services recorded.

34 trains ran on time (defined as less than 1 min late), meaning that 18.3% of those trains that ran were on time.

152 trains were delayed (more than 1 min late).

The average delay (considering all trains that were either on time or late) was 11.5 mins.

December:

145 trains were surveyed (24.6% of all weekday services for the period).

18 trains were cancelled, giving a reliability rating of 87.6%.

26 trains ran on time, meaning that 20.5% of those trains that ran were on time.

101 trains were delayed.

The average delay was 11.3 mins.

A bit of interpretation:

Despite the injection of significant amounts of slack into the winter timetable there has been no improvement whatsoever in punctuality (average delays for October, November and December were all between 11 and 12 mins). Reliability deteriorated markedly, especially in December. For November and December as a whole, more than 80% of trains surveyed were late (more than 1 min behind schedule when observed).

Can I urge observers again to please report the performance of afternoon/evening trains as well as morning ones? You all go off to Oxford, London, wherever in the morning but according to the blog you never come back! And yet there you all are in Charlbury the next morning ready to blog again - curious! How do you all get home - personal jet?

Mon 26 Nov 2007, 21:58

Can people remember to record the afternoon/evening trains that they take as well as the morning ones please? There were lots of records for the morning trains last week but very few for anything that ran after midday. We are now analysing the data each month and sending a summary to FGW but there will be little point doing this in future if the reports are skewed so much towards the morning peak. Thanks.

Sun 11 Nov 2007, 18:54

As promised, a more detailed breakdown of my analysis of the Commuter Blog for September and October. (NB. In all that follows, I have only considered weekday trains as there are currently not enough data for weekend trains. However, keep on blogging at weekends everyone and if there are more recordings in future then I will analyze weekend trains as well.)

September:

104 trains were surveyed (16.8% of all Cotswold Line services for the period).

3 trains were cancelled, giving a reliability rating of 97.1% for those services recorded.

24 trains ran on time (defined as less than 1 min late), meaning that 23.8% of those trains that ran were on time.

77 trains were delayed (more than 1 min late).

The average delay (considering all trains that were either on time or late) was 7.9 mins.

October:

154 trains were surveyed (21.6% of all Cotswold Line services for the period).

4 trains were cancelled, giving a reliability rating of 97.4%.

26 trains ran on time, meaning that 17.3% of those trains that ran were on time.

124 trains were delayed.

The average delay was 11.2 mins.

Susie: I don't have time to do any more on this at present. Are you able to put these figures into a letter, get it signed by some fellow commuters and then forward it to Andrew Haines? I would be tempted to take the line "This is not very good is it and it seems to be getting worse!" A cursory glance at the reports for November suggests that the situation has been pretty atrocious so far this month (big delays reported, including one of more than 2 hours!) and the introduction of the new timetable is only a few weeks away, which is sure to screw things up even further! Thanks. Let me know if you need anything more from me.

Sun 11 Nov 2007, 14:17

Susie: regarding the figures I reported earlier this week, I only considered those services for which reliable data were available in the blog. When I post the expanded analyses (hopefully later today) I will indicate what proportions of all Cotswold Line trains were surveyed for both September and October.

I was also a little bit surprised that the October figures weren't worse than they were. However, the average delay has increased by 38% from 7.9 mins in September to 11.2 mins in October which seems quite significant to me. Think about what this means in real terms - every time you go to catch a Cotswold Line train it is likely to be, on average, 11 mins late. Given that most people arrive at a station 5-10 mins before their train is due to depart, they are liable to spend on average up to 20 mins waiting for that train to arrive. With winter fast approaching, this means a lot of hanging around on cold, wet, draughty platforms, often with little in the way of shelter (think how unappealing Charlbury is when the ticket office is closed!). Anyway, more detail later...

Thu 8 Nov 2007, 22:31

Well! I would really have thought it would have been worse. Did you take the percentage of all the trains that run, regardless of whether people had put times on blog, or only those trains which were noted? But thanks for doing this Derek and would be interested to see fuller analysis.

Thu 8 Nov 2007, 21:32

I completed a preliminary analysis of the October blog reports last night. The headline findings are these:

Reliability (proportion of trains that actually ran) is up slightly from 97.1% in September to 97.4% in October.

The average delay of the trains surveyed rose from 8.0 mins in September to 11.1 mins in October.

There were almost 50% more blog entries for October than September so well done everybody and keep it up for the rest of the year!

I will double-check these figures and hope to get a fuller analysis out by the early part of next week.

To answer John D's query from earlier: I am adopting a "zero-tolerance" approach to punctuality in this analysis, i.e. any train that is delayed by more than a minute is considered to be "late".

Wed 31 Oct 2007, 22:27

Oh come on Ed! FGW are always blaming someone else. I heard the delays this morning were due to broken down trains in Bristol! All I am saying is that this month has been particularly bad. Not all delays have been registered on the blog. We all have one interest in mind, which is to get the trains running on time. Perhaps if we give FGW some statistics on the lateness of their trains, how full they are etc. they may sit up and take note. Apparently, this is one of their most lucrative lines! It's the least they can do. We are the customer - and they should take note.

Wed 31 Oct 2007, 20:35

Surely, Wednesday 31st cannot be used as a typical day. As reported by BBC News and FGW website, over-running Network Rail engineering work at Pangbourne caused delays, diversions and cancellations for most of the morning. If you tried to use today for any meaningful analysis it would undermine the whole argument.

Wed 31 Oct 2007, 11:44

So that Derek can get really bad figures (not difficult) please can everyone who travels on the trains today, enter comments on the commuter blog. This will give a realistic result, especially as the trains have been so bad this month.

Tue 30 Oct 2007, 22:05

Derek

Will you be using the train companies' definitions of 'late' because these don't necessarily match passengers' expectations where 1 minute late is just that! I'd prefer if any late arrival is measured as such...

Tue 30 Oct 2007, 16:04

Sure are, Derek! Happily I have been away on holiday in Florence for a week so have had no train anguish to report of late.

Mon 22 Oct 2007, 17:01

Dear Caroline and Dave: I read your comments just an hour or so after having received a very bad piece of news and I must admit I probably flipped as a result. Sorry to have been rather intemperate. I think the point I was trying to make was that if you are going to criticise someone then it is nice if you try to soften the blow by first giving them at least a tiny dollop of praise! Anyway, lecture over. Hopefully we are all still friends and can continue with the important business of trying to improve the Cotswold Line service for the benefit of all rail passengers.

Mon 22 Oct 2007, 09:38

Sorry, yes the 10 minute rule only applies to high speed services which I presume would be the two morning London bound HSTs and three afternoon return HSTs.

Sun 21 Oct 2007, 14:11

Hi - Many apologies Derek, I didn't mean to sound ungracious about all the the work you've put in. Many thanks for that. It's just that I'm actually really suprised that the reliability is so high, and the delay so low. It just doesn't tally with my own experience. I suspect issue may be that certain trains are just worst than others. I hesitate to suggest more work for you or some other poor soul, but actually tracking the reliability of particular trains might actually reflect which are the least reliable and ones which really get us all hopping mad!!!

Sat 20 Oct 2007, 10:34

I'd also like to thank Derek - and all of us who have been keeping this record on behalf of FGW ! I was not trying to devalue Derek's analysis of the reports, but only to point out that FGW's likely response to that data would be to point out the skewing effect in the reporting.

However I find the result that 12% of all services during the period were reported as delayed (even on a "zero-tolerance" basis) a pretty shocking one.

In future I will endeavour to report on-time services as well as delays experienced.

The more data we gather, the harder to refute the results.

Fri 19 Oct 2007, 23:33

Derek, you are doing us all a great service - keep it up!

The train companies use two different criteria for 'late'; Commuter trains are 'late' if arriving later than 5 minutes at the final destination and I think the former Thames Trains service - our Cotswold Line - fits into this category.

Inter-City trains are 'late' if their arrival at the final destination exceeds 10 minutes over the schedule. Maybe the HST trains fit this category?

Notice that the 'booked time' outward from Paddington to Reading is much shorter than the Reading to Paddington journey. This is because (even in British Rail days) some 5-10 minutes of so-called 'recovery time' was added to the final leg of the journey to increase the likelihood of an 'on time' arrival. Because of the 'final destination' rule, train operators are not/ do not measure lateness at intermediate stations, like Oxford and Charlbury. Probably because it would take an army of clerks to record and monitor such delays!!!

Fri 19 Oct 2007, 22:21

Thanks for your support Susie - at least someone appreciates me! Like you, I thought that rail companies generally define a late train as one that is more than 5 mins late. I tend to adopt a "zero tolerance" approach to train punctuality and in my analysis of the blog I have defined "late" as not being on time, i.e. any delay more than 1 min.

I agree about waiting until the October figures are available. The frequency of blogging has increased recently so keep it up everyone and I'll produce more figures in early November.

Fri 19 Oct 2007, 18:58

First of all, I think we all owe a big thank you to Derek for offering to do this. So thank you Derek, on behalf of all us commuters. I agree that it may be better to wait for October figures, as they are particularly bad. However, we must all try and urge everyone who uses the train to report anything - including those on time - because as Caroline says, the results will be skewed. And with regard to J Norris's post, that a late train is anything after 10 minutes - well! I had always thought it was meant to be 5 minutes, but in my book anything later than the stated timetabled time is late.

Derek, shall we wait until the end of October and then I will try and get some signatures at the station. But well done.

Fri 19 Oct 2007, 15:02

Another issue you might like to take on board Derek is that FGW's definition of "on time" is IIRC within 10 minutues of scheduled arrival. So the average 8 min late is actually 'on time' in FGW-land.

Fri 19 Oct 2007, 14:58

Like Derek, I report all of the trains I take, delayed or otherwise.

Fri 19 Oct 2007, 10:47

"Thanks very much Derek for using your valuable time to analyze the figures for us." (!) Some people are never satisfied...

Why do you assume that most of the reported services are delayed services? For the last six months or so I have reported every FGW train I have witnessed running, be it late, on time, cancelled or whatever. A lot of my reports have been of trains running on time - go back and check my blog entries if you don't believe me!

The Blog for September covers roughly one train in six that stops at Charlbury. Our 76 late-running trains comprise about 12% of all services for this period but this is an utterly meaningless stat because we have no data for the 83% of unreported trains.

The Commuter Blog is only intended to provide a snap-shot of FGW's performance, centring on services to/from Charlbury. No-one is pretending it provides a definitive picture. So far, it looks as if October will provide far more data than September so hopefully a clearer picture will emerge then.

If you want more stats then no-one is stopping you analyzing the figures yourselves! Perhaps you will be able to prove that FGW are ruddy marvellous and that we are all complaining about them for nothing.

Fri 19 Oct 2007, 09:33

Yes indeed - the vast majority of reports are of problems which skews the stats.

Thu 18 Oct 2007, 20:05

We have to assume that most of the services reported on the blog will be delayed services. It would be useful to know what % of all services were reported and what % of all services were then reported late.

Thu 18 Oct 2007, 17:32

I have just concluded a preliminary analysis of September's data from the Commuter Blog. The headline findings are these (NB. I have only analyzed weekday trains as I felt there was insufficient data to do weekend trains as well):

103 services were reported.
Three of these were cancelled, giving a reliability rating of 97.1%.
Of the remaining 100 services, 24 ran on time and 76 were late.
The average delay was exactly 8 minutes.

I hope to be able to produce a slightly more detailed breakdown for September by the start of next week. Susie - if I email it to you in the form of a letter can you get some fellow commuters to sign it and then forward it to Andrew Haines? Alternatively, you might want to wait a couple of weeks and then we could send data for Sept+Oct combined. Let me know what you think.