Ticket Splitting Website - TrainMiser

[Tetchytyke]

I looked from Apsley to Halifax and it was suggesting singles for both the outward and the return, despite the fact they added up to more than the VIA MANCHESTER SVR.

 

[Deerfold]

The search is based on Arrival / Departure times or Departure / Arrival, this is needed to find split tickets efficiently.

The search form maybe needs some more work to make this clearer - I'll look at adding a warning if the inbound dropdown is changed.

I appreciate you need times to find the splits but I don't understand why this can't work just as well with a departure time provided in each direction.

 

[SickyNicky]

It's the same list?

Not necessarily. It might be at the moment, but in the past it hasn't always been.

 

[yorkie]

It's not entirely clear if the routeing "+VIA LONDON" or "NOT VIA LONDON" refers to the routeing group or the station group.

For sure, the routeing "NOT VIA BIRMINGHAM" is intended to mean "Not via Birmingham Stns" (e.g. changing at Aston is OK) rather than "Not via Birmingham Group" however I've discussed this with a couple of people I know, one of whom was denied this ticket from the booking engines, and the other works for LM, between us we concluded that what the booking engines implement is not the same as what the pricing managers believe should be implemented.

Some fares are routed "VIA GLOUCESTER" but the journeys are most obvious via Cheltenham, and it has been argued that the via Gloucester routeing was intended to mean "via Gloucester Group" of which Cheltenham used to be a member.

In some cases station groups and routeing groups are the same, but not in all. There is no "Leeds Stns" but there is a "Leeds Group" for routeing purposes, which permits doubling back between Micklefield & Leeds - which is essential for some journeys on Sundays, or evenings, and very useful during daytimes.

 

[SickyNicky]

It's not entirely clear if the routeing "+VIA LONDON" or "NOT VIA LONDON" refers to the routeing group or the station group.

For sure, the routeing "NOT VIA BIRMINGHAM" is intended to mean "Not via Birmingham Stns" (e.g. changing at Aston is OK) rather than "Not via Birmingham Group" however I've discussed this with a couple of people I know, one of whom was denied this ticket from the booking engines, and the other works for LM, between us we concluded that what the booking engines implement is not the same as what the pricing managers believe should be implemented.

Some fares are routed "VIA GLOUCESTER" but the journeys are most obvious via Cheltenham, and it has been argued that the via Gloucester routeing was intended to mean "via Gloucester Group" of which Cheltenham used to be a member.

In some cases station groups and routeing groups are the same, but not in all. There is no "Leeds Stns" but there is a "Leeds Group" for routeing purposes, which permits doubling back between Micklefield & Leeds - which is essential for some journeys on Sundays, or evenings, and very useful during daytimes.

Ooops. I've got it the wrong way round, but only for via/not via London routes which seem to be a special exception - I've checked the documentation and it is London Terminals that are excluded, not the london group:

If the route descriptor associated with a fare excludes LONDON, then the journey may not pass
through or stop at a London terminal, but may pass through or stop at any other station in the
London Stations list which is not a London terminal.

I will modify my original post. Yorkie makes an excellent point about the distinction between fare groups and routeing groups in via/not via routes.

The problem is that the London Terminals example above is at odds with the routeing guide datafeeds interface specification which says:

GROUP_MKR
Y or N to indicate whether the CRS code represents an individual location (Value=N), or one station in a routeing guide group (Value= Y). If Y, then the whole of the group is included in the route (i.e. the route must include one of the locations in the group (for A or I type entries) or must exclude all the locations in the group (for E type entries))

This is probably what leads to the via Birmingham/Aston example given by Yorkie. It also deals with situations where individual stations are intended rather than whole groups (the GROUP_MKR is set to N in this case).

 

[OwlMan]

I understood that there is a list of Groups for Ticket Routeing purposes included within the timetable data. This list is different from the normal Routeing Groups and from the Ticketing "Stations" groups. A copy is attached from my office records.

Peter

 

[SickyNicky]

I understood that there is a list of Groups for Ticket Routeing purposes included within the timetable data. This list is different from the normal Routeing Groups and from the Ticketing "Stations" groups. A copy is attached from my office records.

Peter

Ah - I can see the file you mean - definition 'G''V' in the master station file. Is this documented anywhere? The public files that describe the data appear to still point towards using the routeing group.

 

[OwlMan]

Ah - I can see the file you mean - definition 'G' in the master station file. Is this documented anywhere? The public files that describe the data appear to still point towards using the routeing group.

I think it is V not G

 

[SickyNicky]

I think it is V not G

Yes - V, sorry. There's no indication in the public documentation what it's for, though, but now that you explain it, it makes sense.

Edit: Yes - when this is properly implemented it will lead to not via Birmingham being valid via Aston, and fix the other anomolies. I just wish they would keep the documentation up to date!

 

[OwlMan]

Yes - V, sorry. There's no indication in the public documentation what it's for, though, but now that you explain it, it makes sense.

I don't know about documentation but that is the list we use at work.

 

[SickyNicky]

I don't know about documentation but that is the list we use at work.

Thanks for that - I imagine it's useful for a number of people and should satisify the problem that TrainMiser has.

I'll stop hijacking your thread now, TrainMiser

 

[CyrusWuff]

One thing missing is the "Boundary Zone" pseudo-stations for use with London area Travelcards, though I'm assuming this is purely for practical reasons as no trains actually stop at same.

I believe this mostly affects longer distance journeys, where it can work out significantly cheaper to do multiple splits (e.g. travelling from London - Liverpool in the evening peak is normally £150.50, but it's possible to get it down to £60.50 by buying Boundary Zone 4 - Wolverton, Wolverton - Rugby, Rugby - Rugeley, Rugeley - Crewe and Crewe - Liverpool).

 

[bnm]

Labour of love TrainMiser, or plans to monetise your split ticketing service in the future as others have done?

 

[kkong]

Does not work for the well known Aberdeen-Stonehaven + Stonehaven-Dundee split for Aberdeen-Dundee.

It offers the standard through fare of £29.80.

 

[TrainMiser]

One thing missing is the "Boundary Zone" pseudo-stations for use with London area Travelcards, though I'm assuming this is purely for practical reasons as no trains actually stop at same.

I believe this mostly affects longer distance journeys, where it can work out significantly cheaper to do multiple splits (e.g. travelling from London - Liverpool in the evening peak is normally £150.50, but it's possible to get it down to £60.50 by buying Boundary Zone 4 - Wolverton, Wolverton - Rugby, Rugby - Rugeley, Rugeley - Crewe and Crewe - Liverpool).

I'll look at adding these - I guess this would be most useful similar to a railcard, i.e. select which Travelcard you have, and then it would use Boundary tickets where available / cheaper?

Does not work for the well known Aberdeen-Stonehaven + Stonehaven-Dundee split for Aberdeen-Dundee.

It offers the standard through fare of £29.80.

The site has a limited number of split points; some smaller stations are skipped to avoid the search taking too long. I'll see if this can be made a bit smarter, i.e. for shorter journeys it could try to split at more stations.

Thanks for that - I imagine it's useful for a number of people and should satisify the problem that TrainMiser has.

I'll stop hijacking your thread now, TrainMiser

Thanks, it's been useful; I think I need to get hold of the "programming rules", though... I'm guessing these are separate from the file description pdfs?

I looked from Apsley to Halifax and it was suggesting singles for both the outward and the return, despite the fact they added up to more than the VIA MANCHESTER SVR.

I'll dig into this in detail tomorrow

Labour of love TrainMiser, or plans to monetise your split ticketing service in the future as others have done?

I'm working on selling tickets directly, it would be good to be able to buy all tickets with one click

 

[185143]

Leeds-Carlisle: one journey routed me via Warrington with a split at SYB and WGN and PRE. The £4.20 SYB-WGN is NOT valid via Warrington as this is a regulated TfGM fare. You have to put up with a 142 on the Kirkby for the £4.20 to be valid.

(Opens can of worms)...Also, you seem annoyingly keen on the Trainline for my liking as I'm SURE you will be told on here!

 

[TrainMiser]

Leeds-Carlisle: one journey routed me via Warrington with a split at SYB and WGN and PRE. The £4.20 SYB-WGN is NOT valid via Warrington as this is a regulated TfGM fare. You have to put up with a 142 on the Kirkby for the £4.20 to be valid.

(Opens can of worms)...Also, you seem annoyingly keen on the Trainline for my liking as I'm SURE you will be told on here!

The £4.20 SYB-WGN is an "Any Permitted" CDR.
In the routeing guide, this is shown as valid on maps GM+XW; this would allow the journey via Warrington.

Also, the National Rail website will show the £4.20 fare if you do a search from SYB-WGN and add "via Warrington".

Again, the links are there to hopefully help support the site. Before long, I'd like to get in place a better process for buying tickets.

 

[SickyNicky]

Agreed. This fare is valid via Warrington. The fare setter is irrelevant. The only revelant thing is the route on the ticket, which is this case is "Any Permitted".

 

[TrainMiser]

I appreciate you need times to find the splits but I don't understand why this can't work just as well with a departure time provided in each direction.

It helps a lot when looking for return journeys.

With a timetable search, going outbound:

• Departure time searches start at the first station and work forwards in time to the second station
• Arrival time searches start at the second station and work backwards in time to the first station

For the return journey, these are reversed:

• Arrival time searches start at the first station and work backwards in time to the second station.
• Departure time searches start at the second station and work forwards in time to the first station

Pairing together the departure and arrival times means that both the outbound and return searches can start at the same station and work towards the other station.
This makes finding splits for returns much easier.

 

[Tetchytyke]

The site has a limited number of split points; some smaller stations are skipped to avoid the search taking too long. I'll see if this can be made a bit smarter, i.e. for shorter journeys it could try to split at more stations.

FWIW I think that is sensible. Trainscanbecheaper is pretty much unusable because it just goes on cheapest, rather than practicality; suggesting a split at Polesworth (as that website once did for me) is daft.

 

[kieron]

I've had a bit of a look at Trainmiser now,

I do really like the way Trainmiser presents different options as a trade-off between speed and cost.

I also like the fact that it gives a result for Wrexham Central-Mouldsworth. It sends me via Rock Ferry (I suppose because that's the closest place it checks for splits), but I'm pleased that it didn't get confused by the lack of through tickets and give up.

I also tried Wrexham Central-Moorfields when I was looking at it. It didn't suggest the split at Rock Ferry, although this would save 20p against the through ticket it actually suggested. I don't know why.

Ah - I can see the file you mean - definition 'G''V' in the master station file. Is this documented anywhere? The public files that describe the data appear to still point towards using the routeing group.

My reading of the timetable feed specification is that those lines are just to do with the fare check, to say something like "if fare check fails for MRF, don't try with LIV or LVC". As there's no guidance on which station to try first, it should have no effect on the results.

I suspect this data may be obsolescent, anyway, as it includes stations such as West Ham which used to be routeing points, but not ones like Weston Super Mare, which became a routeing point last year.

I am a bit perplexed by "EARLESTOWN (CHESHIRE)", though. I know it's less than a mile from the Warrington Borough Council boundary by foot, but that's not how it works.

FWIW I think that is sensible. Trainscanbecheaper is pretty much unusable because it just goes on cheapest, rather than practicality; suggesting a split at Polesworth (as that website once did for me) is daft.

TCBC's problem was that it didn't consider timetables at all, but only permitted routes (I believe it now also suffers from using an out-of-date version of the routeing guide). You don't have to heavily restrict the range of split points to avoid that problem, as looking at the timetable and only giving splits which aren't too slow should be enough. Preferring to split at better served stations is a good idea, but excluding so many stations altogether is bound to hurt the results. It does save time, though.

 

[34D]

I'll look at adding these - I guess this would be most useful similar to a railcard, i.e. select which Travelcard you have, and then it would use Boundary tickets where available / cheaper?

Not just this. There may be some journeys where buying a zone 1-4 travelcard and a bz4 to wherever are cheaper - though these should be the exception rather than the rule.

Maybe a zone 2-6 travelcard may be more likely to feature?

Can you add the functionality for someone to say yes/no to "relevant season ticket held"?

 

[yorkie]

TCBC's problem was that it didn't consider timetables at all, but only permitted routes (I believe it now also suffers from using an out-of-date version of the routeing guide). You don't have to heavily restrict the range of split points to avoid that problem, as looking at the timetable and only giving splits which aren't too slow should be enough. Preferring to split at better served stations is a good idea, but excluding so many stations altogether is bound to hurt the results. It does save time, though.

TCBC had a 'cheapest' mode, and this did not consider the timetable. It did what it was designed to do!

TCBC did have a timetable search mode, which evolved into Trainsplit.
--- old post above --- --- new post below ---

It helps a lot when looking for return journeys.

With a timetable search, going outbound:

• Departure time searches start at the first station and work forwards in time to the second station
• Arrival time searches start at the second station and work backwards in time to the first station

For the return journey, these are reversed:

• Arrival time searches start at the first station and work backwards in time to the second station.
• Departure time searches start at the second station and work forwards in time to the first station

Pairing together the departure and arrival times means that both the outbound and return searches can start at the same station and work towards the other station.
This makes finding splits for returns much easier.

Yep, got it, but I think we can hide this to the user, by asking for whatever users want and then setting the algorithm to do as described above.

For example, on Trainsplit, I choose departure times for both my outward and return journeys. But before I press submit, the arrival times have already been estimated for me. So perhaps the estimated arrival time of the return journey can be used by the algorithm above to work in the way described?]

Hopefully that makes sense. Is it possible?

 

[34D]

trainmiser I tried your site on this thread but it returned nothing but the £124.50 anytime single for each way.

 

[yorkie]

Can you add the functionality for someone to say yes/no to "relevant season ticket held"?

trainmiser I tried your site on this thread but it returned nothing but the £124.50 anytime single for each way.

I'm not sure if websites will be allowed to sell tickets which are to extend a journey covered in part by a Season ticket.

And even if they are, it'll be a niche market, so I expect it would be low down on the list of priorities.

 

[TrainMiser]

Yep, got it, but I think we can hide this to the user, by asking for whatever users want and then setting the algorithm to do as described above.

For example, on Trainsplit, I choose departure times for both my outward and return journeys. But before I press submit, the arrival times have already been estimated for me. So perhaps the estimated arrival time of the return journey can be used by the algorithm above to work in the way described?

I think this could cause more confusion.

The journeys returned often have different lengths, e.g. suppose the fastest journey is 4 hours, and the slowest / cheapest journey is 6 hours.

If someone asks for departures from 6pm, this would be converted to a 10pm arrival (based on the fastest journey), which would then show departures starting at 4pm...

In practice, I don't think the restriction is too much of an issue, as most return journeys are either:

• Journeys to work / an interview / a dentist's appointment (Arrival / Departure works best), or
• Journeys for a day out (Departure / Arrival works best)

I think the cases where a Departure / Departure search is best are fairly uncommon.
--- old post above --- --- new post below ---

Not just this. There may be some journeys where buying a zone 1-4 travelcard and a bz4 to wherever are cheaper - though these should be the exception rather than the rule.

Do you know of any examples of this?
--- old post above --- --- new post below ---

trainmiser I tried your site on this thread but it returned nothing but the £124.50 anytime single for each way.

Do you know which search you ran?

It looks as though there are two results returned for this, the fastest is £249 total, the second involves changing onto FCC at Peterborough, and is £111.50 total.
--- old post above --- --- new post below ---

I also tried Wrexham Central-Moorfields when I was looking at it. It didn't suggest the split at Rock Ferry, although this would save 20p against the through ticket it actually suggested. I don't know why.

The fastest route for the search has a change at Bidston. Travelling via Rock Ferry would add time to the journey, and wouldn't be worth it for the saving.

 

[Hadders]

Do you know of any examples of this?

There are several examples from the Great Northern route where making use of FCC's super off-peak day travelcards at weekends and a BZ6 fare is cheaper than the through ticket price.

 

[nicobobinus]

Simple and clear feel, looks good. Behaves fine on my imac running Chrome.

Would like to see a railcard option to show network splits and the like, but know that would need a fair bit more work!

 

[Deerfold]

Just had an odd one - was looking up Haddenham and Thame Parkway to London Marylebone for this evening, arriving 1800 and leaving 2300.

Coming back it puts me on the 0005 train, ignoring the direct 2320 which also serves the same split point at Princes Risborough.

The split's only a saving of 30p so I may not bother with it anyway.

 

[SickyNicky]

At a guess, I would say this is related to the time of the return journey. The 2320 doesn't leave Princes Risborough until 0007 the next day, so maybe it's disallowing the off-peak day return because the return's on a different day?

Tickets are valid, of course, until 0430 the next day.