TransXChange to GTFS - Bus, Coach, Tram, Metro and Ferry

Malcolm Morgan

2026-07-15

Source: vignettes/transxchange.Rmd

This vignette explains how to convert TransXChange timetables into GTFS, the difference between the two main sources of TransXChange data (Traveline and the Bus Open Data Service), and the pitfalls to watch out for.

Background: what is TransXChange?

TransXChange (often abbreviated TXC) is the UK Department for Transport’s XML standard for describing bus, coach, tram, metro and ferry timetables. It is the format that operators must use to register services with the Traffic Commissioners, so almost every scheduled road-based service in Great Britain is described in TransXChange at some point in its life.

Because it was designed for registration rather than for journey planning it has some awkward properties:

  • It is verbose XML. A single medium-sized operator can run to tens of thousands of files.
  • One file describes one service, and the same real-world journey is often spread across many files (and re-uploaded every time the timetable changes - see Duplicate and superseded files below).
  • It does not contain stop coordinates. Stops are referenced only by their ATCO code, and you must look the location up in the separate NaPTAN database (see Stop locations - the NaPTAN).
  • Operating patterns are described with rules (“every weekday except bank holidays”) rather than explicit dates, so a calendar has to be computed.

GTFS, by contrast, is a flat set of CSV files designed to be consumed directly by journey planners such as OpenTripPlanner, R5 and Google Maps. Converting TransXChange to GTFS makes the data usable by that wide ecosystem of tools.

Where to get TransXChange data

There are two national sources. They contain broadly the same underlying registrations but differ in packaging, freshness and coverage.

1. Traveline National Dataset (TNDS)

Traveline publishes the Traveline National Dataset (TNDS). You must apply for free access to their FTP server. The data arrives as zipped folders, one per region of Great Britain (for example EA.zip for East Anglia, S.zip for Scotland).

  • Pros: organised cleanly by region; includes coach, some ferry and light rail; a long-established, relatively stable feed.
  • Cons: each region is a separate download; updated less frequently than BODS; being progressively superseded by BODS as the DfT’s preferred source.

2. Bus Open Data Service (BODS)

The Bus Open Data Service (BODS) is the DfT’s newer, statutory feed. Since 2020 bus operators in England are legally required to publish their timetables here, so it is the most complete and up-to-date source for English buses. You can download TransXChange directly, or download a ready-made national GTFS file produced by ITO World’s own converter.

  • Pros: most current data; legally mandated so very complete for England; offers a ready-made GTFS download.
  • Cons: English buses only (Scotland and Wales still rely on TNDS for full coverage); the bulk “change” archives accumulate many superseded versions of the same service, which will inflate trip counts if converted naively (see below); tram, metro and ferry are patchy.

Which should I use?

  • If you just want current English bus GTFS and are not fussy about the details of conversion, download the ready-made GTFS from BODS - it is the least effort.
  • If you need tram, metro, ferry or coach, need Scotland or Wales, want a second independent conversion for comparison, or need historical timetables, use UK2GTFS. The package adds extra data checks and supports the non-bus modes that the ready-made BODS GTFS omits.

Converting TransXChange to GTFS

The workhorse is transxchange2gtfs(). In its simplest form:

library(UK2GTFS)

path_in <- "EA.zip" # a TNDS region zip, or a folder of .xml files
gtfs <- transxchange2gtfs(path_in = path_in, ncores = 3)

Key arguments:

  • path_in - a single zip folder, or a character vector of paths to .xml files.
  • ncores - if > 1, multi-core processing is used to speed things up. Always leave one core free for the operating system.
  • try_mode - if TRUE (the default) files that fail to convert are skipped rather than aborting the whole run. This is robust, but be aware the result may be missing some routes - check the messages.
  • cal / naptan - the bank-holiday calendar and stop locations. By default these are fetched for you with get_bank_holidays() and get_naptan(). If you are converting many files it is much faster to fetch them once and pass them in (see below).
  • scotland - "auto", "yes" or "no". Controls whether Scottish bank holidays are used. "auto" treats a file ending in S.zip as Scottish.
  • filter_duplicate_files - see Duplicate and superseded files.

Once converted, save the GTFS to disk:

gtfs_write(gtfs, folder = "C:/GTFS", name = "gtfs_EA")

This writes C:/GTFS/gtfs_EA.zip.

Converting many regions efficiently

get_naptan() and get_bank_holidays() download data from the internet. When looping over several regional zips, fetch them once and reuse them:

library(UK2GTFS)

naptan <- get_naptan()
cal    <- get_bank_holidays()

files <- list.files("path/to/tnds", pattern = "\\.zip$", full.names = TRUE)

for (f in files) {
  message(f)
  gtfs <- transxchange2gtfs(
    path_in = f,
    cal     = cal,
    naptan  = naptan,
    ncores  = 3
  )
  name_out <- gsub(".zip", "", basename(f), fixed = TRUE)
  gtfs_write(gtfs, folder = "path/to/output", name = name_out)
}

How the conversion works

transxchange2gtfs() is a convenience wrapper around several lower-level steps, which you can also call individually:

  1. transxchange_import() reads each XML file into R data frames.
  2. transxchange_export() turns those into a GTFS object.
  3. gtfs_merge() combines the per-file GTFS objects into one.
  4. gtfs_write() saves the result.

Each TransXChange file is converted into its own small GTFS object and these are then merged. Understanding this matters because the merge is where problems tend to surface.

Stop locations - the NaPTAN

TransXChange files reference stops only by their ATCO code; they do not contain coordinates. UK2GTFS therefore looks each stop up in the NaPTAN (National Public Transport Access Nodes), the DfT’s database of every public transport stop in Great Britain.

get_naptan() always downloads the latest copy. The NaPTAN is not perfect - it misses some stops and has wrong locations for others - so UK2GTFS ships two internal correction datasets, naptan_missing (stops absent from the NaPTAN) and naptan_replace (better coordinates for known-bad stops), which are applied automatically. Contributions of new or corrected stops are welcome via the UK2GTFS-data repository.

Merging problems

Because each file is converted separately and then merged, the merge can fail on small inconsistencies between files - most commonly when the same operator is spelt two different ways (for example "Yorkshire Coastliner Ltd" versus "Yorkshire Coastliner"), which looks like two agencies sharing one ID.

If transxchange2gtfs() cannot merge cleanly it returns a list of GTFS objects instead of a single object. You can then merge them manually, forcing past the conflict:

gtfs <- gtfs_merge(gtfs_list, force = TRUE)

force = TRUE keeps the first instance of each duplicated ID, which is almost always the correct behaviour. You can also set force_merge = TRUE in the original transxchange2gtfs() call to do this automatically.

Duplicate and superseded files

This is the single most important thing to understand when working with bulk archives (especially the BODS change archives).

Every time an operator revises a timetable they upload a new TransXChange file for the same ServiceCode, but the older, superseded files usually remain in the archive and often still declare an open-ended operating period. If you convert all of them, the same physical bus journey appears once per file version, so counting trips on a given date over-estimates service - in one test archive a single route appeared five times.

A normal single download of current data does not have this problem (each service appears once), so you only need to worry about this for accumulating archives.

Set filter_duplicate_files = TRUE to keep only the operative version of each service:

gtfs <- transxchange2gtfs(
  path_in                = "bods_archive.zip",
  filter_duplicate_files = TRUE,
  filter_date            = as.Date("2024-06-01"), # a date inside the period you care about
  ncores                 = 3
)

For historical analysis, set filter_date to a date within the period you are studying - the filter keeps the version that was operative on that date. Under the hood this calls txc_filter_files(), which you can also run directly on a vector of file paths; see ?txc_filter_files for the exact rules and their one documented limitation (trip counts can still double-count after the next scheduled timetable change).

Comparison with the ready-made BODS GTFS

UK2GTFS output has been validated against BODS’s own (ITO World) GTFS. Where both feeds cover the same route, agreement is high - around 80% of shared operator/line pairs had identical daily trip counts and matching departure times. The remaining differences are mostly about which input files are included (superseded revisions, out-of-region services, and coach/ferry datasets that live outside the TransXChange archive) rather than the conversion itself. In short: both are reasonable, and having an independent open-source converter is useful precisely because TransXChange-to-GTFS involves interpretation.

Next steps

Once you have a GTFS file you will usually want to clean and check it. See the Working with GTFS files vignette for gtfs_clean(), gtfs_validate_internal(), gtfs_clip() and the analysis functions.