I am looking for a way to automatically compute ridership numbers for my country’s intercity railway services. I did a first attempt at an excel sheet, which spits out numbers of travellers between one of 34 agglomerations in Kojo. The next step would be to allocate those numbers on route segments to visualise accurate cross sections of passenger flows on the system, but I’m not sure I know how to get their from a technical stand point. Advise is very welcome.
Here’s the link, and there’s a preview down below.
I just want to put these 5 notes here for everyone to read who wants to have a look at this or might fool around with it themselves:
1) “Spawn” and “destination” does not imply a trip from spawn A to destination B. Rather, cell J7 for example says: There are 1934 trips taking place per day between Kwaengdō and the Finkyáse agglomeration, where the trip maker is living in Kwaengdō and temporarily staying in Finkyáse. Combine that with the 1479 trips where a Finkyásenian is visiting their grandmother in Kwaengdō, and you get a total of 3143 trips per day between the two cities, or roughly 1570 per direction.
2) A number of riders “gets lost” in the formulars, as the attraction shares never add up to exactly 100%. For example journeys from Pyingshum to Pyingshum are obviously not taken into account; that means the number of trips per person and year should be set slightly higher than the actual value.
3) While the attraction share can kinda compensate for this, one should keep in mind that in the real world for example a young student city will spawn much more trips per inhabitant than let’s say a former industrial town with much poverty. This model assumes every city spawns the same amount of trips per inhabitant, and just their destinations are determined by the attractiveness of their destination city.
4) Neither the distance, nor the ease of travelling (by train) between two cities is taken into account. I feel like partly this cancels out with other effects, for example really close cities might have a higher travel demand to each other, but on the other side regional trains might become a more attractive alternative on those relations. But obviously the 27+45 travellers between Shangmē/Chin-Jōrin and Oreppyo would much rather take the bus for a quick 1 1/2 hour trip than taking a 150 km detour via Pyingshum.
5) This excel sheet is ignoring the (quite significant) cross-border traffic, for example to Ataraxie-Ville.
Thank you for your interest, and hopefully your feedback! Cheers,
Comment from Rustem Pasha on 7 October 2017 at 18:39
The method is pretty close to IRL methods of traffic prognostication but you should make everything to minimalize problem with missing trips. It is not a problem with people who travel just for travel or time spending, thise travels are usually less than 0.02% of all travels. Much more than it you have to have as a seats reserve in trains because number of journeys taken by people vary in the year (for example for polish railways first day of winter holidays is really hard, the same at New Years Eve).
Second thing is that attraction multiplier is too exaggerated. About 50% of it should vary between 0.90 and 1.10, in some bigger cities it could high to ~1.20-1.30 and 1.50 in the capital or other important bussiness center (you rather don’t have something like that, because Pyingshum aglomeration is much bigger than any other city in Kojo).
Third is problems with passengers from other countries. IRL they are added at (even fictional, created just for prognostication) train stops at the border.
Fourth is that your algoritm don’t consider rural population. They rather use IC trains but less frequently than people from the cities (attraction mutiplier ~0.70-0.80). How to consider them? Well, it’s a lot of work. All of the country should be divided on zones depending on the station which people from some places use (time travel to it is the shortest) and then you should evaluate how much people live in each zone and then add this zones and population to the Excel. If this is too lot of work you can simply multiply every number of trips by 40000000/32620000 but the results will be less accurate in this case.
Anyway great work. I will probably never motivate myself to do something like that.
Comment from Leowezy on 7 October 2017 at 21:26
Hey Rustem, thanks for your comments!
I’m aware that ridership will vary heavily throughout the year; the numbers are not intended to be used as actual seat capacities on the trains, but rather to give a vague idea whether trains should run every 30 min or only every 2 hours. And since I’m a horrible infrastructure enthusiast, I’ll be setting the seat capacity and therefore train frequency generously high compared to the actual numbers coming out of this spreadsheet ;)
I’d like to disagree on your statement about the attraction multiplier, and I think it might even be sensible to set it even higher/lower in some places. Looking at tourism cities like Ro and Tsuyenji, they easily attract those numbers in domestic tourists, in relation to their relatively small local population. With places like Kari on the other hand, there’s really nothing special going on there that would encourage people from other cities to go there unless they have family roots or alike there, so 0.70 might still be too high. But it’s definitely a difficult question, and depends mostly on how I picture every individual city in my mind.
In an earlier version I actually had Ataraxie-Ville (meaning to include the other stops in central Ataraxia as well) as a separate node; I decided against it though, to avoid overlapping with Dono’s competences; Also the attraction multiplier here will depend heavily on how close social and business links are between our two countries, and that’s not 100% fixed yet.
The algorithm does consider rural population, you’ll notice that the population given to each node is considerably higher than the city’s “proper” population, depending on how many people I guessed there would be living in the surroundings of each IC stop. There are still about 8 million people missing though, which I attribute to the fact that I have not yet finished the human geography in most areas in the country.
Comment from Sarepava on 7 October 2017 at 22:59
One weakness of this method, as I understand it, is that it does not present a clear figure for the ridership per journey stage. If we look at the service between Pyingshum and Kippa, this is 13750, but does that include the 1250 going only as far as Leshfyomi-sul, or is the total passengers for this stage of the journey 15000? I think this would be more useful to display a daily total for each section of the network between stations, not least because for operational reasons it might turn out that the complete 300km route gets one train every two hours, but the first 60km gets three in between because demand is much higher. This feeds into the map because additional platforms and turnback facilities might have to be built, the stations get more facilities etc. Within Pyingshum the journeys between terminus stations via your ‘S-Bahn’ are likely to see as many passengers as long distance routes (so it’s worth getting some figures for these), which creates a loop whereby demand for the IC services puts pressure on the S-Bahn, which has to upgrade to handle the traffic. This is exactly the situation with the imminent Crossrail in London which is an attempt to take medium-distance passengers off the Underground by opening up direct services from mainline to mainline via a central tunnel. In the case of Kojo, all three Dyanchezi already have this ‘dive underground’ facility, so in future this might need expanding to meet demand.
On a wider point, I think such calculations as this can be very helpful tools directly related to mid-level detailing (ie buildings) of our settlements. If there exists a formula to calculate birth rate from population density, this gives an indication of how many schools are required in an area. How many restaurants and bars can be economically supported by a given population? How many supermarkets, and so on.
Comment from Leowezy on 8 October 2017 at 09:07
To simply answer your question, no the numbers only indicate ridership that originates/terminates in the respective nodes. So I guess I’d have to assign route segments to every one of those relations ([34*34-34]/2=561), and then add the ridership of all relations for each relation used by those; then I could generate numbers alike “on the tracks between Leshfyomi-sul and Kippa there travel xxxx people per day in both directions”.
I found that estimating passenger numbers for my railway stations in Pyingshum is actually much more tricky than I thought; it’s one thing to estimate how many people will arrive by terminating IC, CC and regional trains, but how many of them are walking to work from there, taking the metro or the Papache? I tried to guesstimate those numbers on my wiki articles for Aku-Dyanchezi, Limbē-Dyanchezi and Kibō-Dyanchezi.
I really like your last idea as well; I must say I haven’t mapped any city to that level of deatail yet, but I’m especially in negligence of educational amenities. I might try to focus on those aspects more in the future.
Comment from Rustem Pasha on 8 October 2017 at 10:42
@Leowezy, I agree that at this stage of development of the country we can’t evaluate number of trips very accurately but you asked what people think about your spreadsheet, s I answered how it could be improved.
Still I understand what you think about the attraction multiplier but the values which I mentioned are derived straight from RL world values. Generally I have seen the multiplier bigger than 2 only once (it is used IRL vastly for prognostication after population extrapolation, on areas where the traffic research hasn’t been done or just for future sociological analysys). It was about 2.12 in one neighobourhood in such big city. In this neighbourhood most people were young adults employed adults so it is understandable thing. Stil the whole city had multiplier of about 1.43.
This traffic research was interesting because it divided travelers on few categories:
- young people who still taking an education - multiplier 1.84
- employed adults - 2.26
- unemployed adults - 0.65
- retirees - 0.43
Finally I want to write about lacking of information about this 8 millions of people and foreign travels. If you don’t have any idea how the rural population will be placed they should be placed equally in all unplanned region. Even if you make a mistake it will be probably no more than 30%. Now you have a 100% so it is rather profitable. The same with foreigners. They are in the system but you don’t know how much are they. Initially you can prognosticate this number. If you make mistake and, let’s say, write 10000 and it will be 7000 or 15000, your prognostication will be still closer to the truth than 0.